Professional Documents
Culture Documents
DOCTEUR
en
par
Rowshni JABEEN
DOCTORAT DE L’UNIVERSITÉ DE LILLE DELIVRÉ PAR IMT NORD EUROPE
Laboratoire d’accueil
Centre d’Enseignement, de Recherche et d’Innovation (CERI)
Matériaux et Procédés de IMT Nord Europe
Ecole Doctorale SMRE 104 (U. Lille, U. Artois, ULCO, UPHF, Centrale Lille,
Chimie Lille, IMT Nord Europe)
Dedicated to
Waheed, Jyothi, Reshma and Nanda.
Acknowledgements
I begin by thanking my thesis supervisor Prof. Chung Hae Park for all your guidance and
your immense patience during my time at IMT. Thank you to the pillars of this thesis,
Benoı̂t and André for all the encouragement and providing great knowledge for this thesis.
It has been a great pleasure working with you.
I express my deepest gratitude to the members of the jury, Prof. Rocha Da Silva, Dr.
Chabert, Prof. Desplentere and Dr. Laure for agreeing to be a part of this dissertation
and acknowledging my work.
Special thanks go to the project of INTERREG ATHENS for funding this work, and
to the team, especially, Leila Bonnaud, Frederik Desplentere and Sofie Verstraete for the
collaborations and discussions during the project.
Thank you to my colleagues Xavier and Patrice, who have always come to the rescue
when machines pulled tantrums and for being my unofficial french language teachers.
My friends at Douai made my life better than I could ever imagine. Your company
kept me strong and I always knew you guys were there for me. Anurag, I can write a new
thesis only to describe you and your influence on my life. Thank you for being my ‘know-
it-all’ friend. Amulya, you have a special place in my heart. We vibe! Thank you for being
my ‘aww-so-caring’ friend. Aniket, your stories and your experiences inspire me. Thank
you for being my ‘this-guy-is-so-amazing’ friend. Abhilash (a.k.a Amulya’s husband), you
are my ‘never-say-I-dont-know’ friend, you taught me how to show confidence even if you
don’t know the answer. Thank you Keerthi for being my ‘beshht’ friend.
A huge shout out to my friends who have accompanied me through the thick and thin
in life, Abhilash, Pasha, Uthkarsha, Rinky, Nawaz, Chandu, and Chaitanya. My new
family in Europe, Sneha and Kaushik, thank you for giving me great moments and food
that I will always cherish.
My heartfelt gratitude to my parents, Waheed and Jyothi, who are the reason for who
I am today and for always being proud of me. You both are my true inspiration. Reshma,
you are the best sister one can ever have. Thank you for being hopeful and believing that
I am here for you. Finally, my sweetheart husband, Nanda, who stood by me in all my
emotions during this journey. Thank you Nanda for showering all the love and support
and not quitting to believe in me.
Contents
Introduction 1
v
Contents
vi
Contents
Bibliography 103
vii
List of Figures
2.1 Grid used in algorithm showing (a) inclusion interaction (b) inclusion cen-
tres and periodicity [88]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.2 Imposition of obstruction to every ray path. . . . . . . . . . . . . . . . . . 30
2.3 Definition of orientation of unit vector P with angles θ and ϕ. . . . . . . . 31
2.4 Orientation distribution of fibre shown in a unit spheres, generated from
fibre orientation tensor with components a11 = 1 and a22 = a33 = 0. (a)
second-order ODF (b) fourth order ODF. . . . . . . . . . . . . . . . . . . . 32
2.5 Algorithm to generate 3D numerical geometries to perform ray tracing
simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.6 Method verification to use fibre orientation tensors to define orientation
distribution function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.7 Transmission of a laser beam through two different media. . . . . . . . . . 36
2.8 Ray tracing simulation at micro scale. . . . . . . . . . . . . . . . . . . . . . 37
ix
List of Figures
x
List of Figures
4.1 Inflience of physical parameters on weld line formation during laser trans-
mission welding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.2 Rectangular elements representing surface roughness before and after ap-
plying clamping pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.3 Representation of molecular chain diffusion in region c after compression
of unit cell b to achieve an intimate contact [130] . . . . . . . . . . . . . . 78
4.4 Stages of interdiffusion of polymer chains during laser transmission welding 79
4.5 Diffusion of molecular chains according to Reptation theory. . . . . . . . . 81
4.6 Coefficient of interdiffusion at the weld interface as a function of temperature. 84
4.7 Contour plot of temperature with respect to laser speed and power . . . . 87
4.8 Lap-joint setup for laser transmission welding . . . . . . . . . . . . . . . . 88
4.9 Degree of healing vs. melt time for weld speed of 7mm/s . . . . . . . . . . 89
4.10 Degree of healing with respect to melt time for all cases of laser movement
speed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.11 Exponential plot of the calculated reptation time as a function of maximum
temperature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.12 Weld line discretized in five parts. . . . . . . . . . . . . . . . . . . . . . . . 91
4.13 Temperature profile measured at five locations of weld line at interface of
flax fibre reinforced PLA composite and pure PLA plates. . . . . . . . . . . 92
4.14 Degree of healing determined by interpolating melt time t0.25
m,1 with heal
time t0.25
h,1 at location 1 on weld line. . . . . . . . . . . . . . . . . . . . . . . 93
xi
List of Figures
xii
List of Tables
4.1 Welding process parameters for lap shear specimens and HAZ temperature
measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.2 Experimental data to calculate reptation time of PLA at various tempera-
tures at interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.3 Comparison of weld strength predicted experimentally and numerically . . 96
xiii
Introduction
This thesis is part of INTERREG Europe. The project name is ATHENS, which is an
abbreviation to Assembly Technologies for Hybrid structurEs of Natural fibre
compositeS. ATHENS is an integrated cross-border collaboration between the Belgian
regions of Flanders and Wallonia, and France, aimed at the development of various tech-
niques for assembling bio-composites. This project focuses on two assembly techniques
suitable for bio-composites, namely laser transmission welding and adhesive bonding, and
addresses related challenges.
1
The project partners include KU Leuven, IMT Nord Europe, Materia NOVA among
others.
The project goals can be construed in three parts:
IMT Nord Europe carried out a module to model and optimise laser transmission weld-
ing of bio-composites. The objective of this module was to optimize assembly technologies
for welding of simple structures produced by the other partner KU Leuven. The basic
characteristics of the assembly process were identified and compared with the performance
expected by the industrial assembly process.
For the joining of thermoplastic composites, the samples were preprocessed according
to current industrial best practices (pre-treatment and application). The model of a
welding process of composite materials consisting of a thermoplastic matrix and short
natural fibres was developed based on IMT Nord Europe’s previous works on continuous
glass fibre reinforced composite materials.
2
Outline of the dissertation
3
Chapter 1
5
in different parts of the automobile. To be competitive, car manufacturers need to increase
their responsiveness to market demand fluctuations and variability adaptive to complex
assembly systems [4]. Future automotive production systems have to handle a variety of
materials and products of various customized geometries [5]. For some products, joining
is required due to the limitation of manufacturing complex parts with available resources.
However, the path to manufacture composite parts comes with its own concerns. One
of these is, developing a fast and reliable assembly of composite parts. The capability
of thermoplastics to melt and consolidate, welding techniques can be considered here as
an added advantage. Welding of two thermoplastic composite parts requires a physical
contact and sufficient amount of energy at the interface of the substrates to melt the local
thermoplastic volume [6].
Many welding techniques are available to join thermoplastics, such as ultrasonic weld-
ing, linear vibration welding, hot-tool welding, resistance implant welding and dielectric
welding [7, 8, 9, 10]. Welding techniques such as, vibration welding, ultrasonic welding,
friction welding, and spin welding uses mechanical movement to generate heat while tech-
niques such as infrared welding, induction welding and microwave welding employ direct
electromagnetism to induce heat at the interface.
Laser transmission welding (LTW) is a popular joining process for thermoplastic parts
not just in automotive, but also in medical and electronic sectors, especially when the
control of heat source location, size and magnitude is critical [11]. Every welding process
inherits its own merits and limitations, and are chosen based on the assembly requirements
such as joint configuration, cost, speed etc. Laser transmission welding is one of the
prominent welding methods that employ external heat source.
Advantages of LTW [12]:
6
Laser transmission welding of bio-sourced thermoplastic composites
5. No particulate generation
7. No surface damage
Disadvantages of LTW:
3. Part clamping must be designed carefully to ensure contact at the whole of the joint
area during welding
4. Laser absorbing material must be added to one of the plastics or at the joint surface.
The transparent part must transmit the laser radiation efficiently. This can limit
the thickness of the top substrate when welding plastics with low transmission (e.g.
PEEK or some reinforced plastics)
7
1.1. Working principle of LTW
Laser beam
Laser direction
Transparent
part
Weld zone
Clamping
pressure
Absorbent part
the absorbent part melts locally. At this weld interface, the transparent part also melts
locally due to heat conduction. A bond is formed upon solidification. Typically, the
materials to be welded are fixed and clamped together, while the laser source moves. The
procedure is commonly performed on pure thermoplastics with no reinforcements in the
transparent part. This process cannot be performed with thermosets as they cannot be
re-fused [15].
Figure 1.2 shows the various steps of LTW. There are four common stages seen during
LTW process:
• Positioning of the thermoplastic parts that are to be welded. The two parts are
placed in close contact such that the transparent part faces the laser source.
• Transmission of laser beam through the transparent medium and its absorption at
the surface of the absorbent medium where the materials melt locally. The heating
process may vary based on the part geometry and thermal properties.
8
Laser transmission welding of bio-sourced thermoplastic composites
molecular entanglements.
• Cooling or solidification of the interface resulting in the freezing of the newly formed
molecular entanglements and thus forming a weld.
Laser rays
Transfer of energy
through transparent
medium.
Energy absorbed in
absorbent medium.
Polymer melts at
interface
– heat propagates due
to conduction.
Solidification
LTW can be performed only between compatible thermoplastics that have similar melting
temperature profiles, good miscibility and similar melt viscosity. An advantage of LTW is
that it is performed rapidly without melting large volume of the plastic part. Under such
conditions, the flexibility to combine different polymers increases. In [12], the authors
have systematically listed combinations of thermoplastics suited for LTW. In any case,
one of the thermoplastics chosen for the LTW process should ensure laser transmission
and the other, laser absorption[16].
9
1.2. Parameters affecting laser transmission welding
When fibres are included in the transparent part, it becomes semi-transparent and diffu-
sive. The quality of the weld depends on the interface weld temperature which is impacted
by the optical and thermal properties of the composite parts and the welding parameters
[17]. During its propagation in a heterogeneous medium, like fibre reinforced composites,
the laser beam undergoes reflection and/or refraction at every fibre-matrix interface in
their path. This leads to the scatter of laser rays before reaching the weld interface caus-
ing a reduction in laser intensity, and hence attenuation in energy reaching the interface
[18]. It has been reported that the temperature distribution at the weld seam strongly
depends on the local fibre orientation [19].
When the laser radiation reaches the surface of a material, its energy can partly be
reflected, or be absorbed. In case of a semi-transparent material, the laser intensity
can also be diffused by the constituents of the material and finally, a part of the initial
radiation can pass through the volume of the material.
σ = f (P, T, t) (1.1)
The temperature (T ) in the weld area should be above glass transition temperature (for
amorphous polymers) or melting temperature (for semi-crystalline polymers) of the in-
volved materials to reduce local viscosity and allow inter-molecular inter-diffusion of poly-
mer chains [22]. The materials conduct heat efficiently and diffuse polymer chains while
being in an intimate contact with each other. This is achieved with clamping pressure (P )
mechanism. Time (t) is evidently a very important parameter that affects the formation
10
Laser transmission welding of bio-sourced thermoplastic composites
of a good weld.
Overall, development of appropriate temperature at the weld interface is of highest
interest in LTW. The temperature field at the weld interface is influenced by welding
process parameters such as welding speed, laser power, laser beam characteristics, optical
properties of the materials and lastly, thermal properties of the materials.
The laser beam interaction with the materials depends on factors such as:
• Laser beam intensity, power, size, shape of laser spot, laser beam quality.
• Laser beam speed of movement during welding process (in case of moving laser
welding machines).
Commonly used laser types for welding thermoplastics are Nd:YAG, Diode, Fibre and
CO2 [14]. Laser type influences the quality of the weld as the available lasers differ in
wavelength, spot precision and efficiency. Figure 1.3 shows the laser emission wavelengths
positioned in the electromagnetic spectrum.
CO2 lasers
The first laser to be used in welding of thermoplastics was the CO2 laser in 1970 [24].
The CO2 laser radiation (wavelength of 10.6 µm) is promptly absorbed at the interface
11
1.2. Parameters affecting laser transmission welding
of the thermoplastic materials enabling the joining process. In CO2 lasers, the beam is
commonly guided from the source to the workplace through mirrors.
Nd:YAG lasers
Nd:YAG lasers are solid state lasers with wavelength of 1064 nm. It is widely used in
material processing industry. An optical fibre is usually used to transfer laser to the weld
area. For high power requirements, it is feasible to combine multiple Nd:YAG laser beams.
These lasers are well suited for high precision welding as the beam spot is relatively small.
Fibre lasers
Fibre lasers have a wavelength range of 1000-2100 nm. In the field of materials processing,
much interest has focused on wavelengths around 1,100 nm to provide a replacement for
Nd:YAG lasers, with equivalent beam quality, but greater efficiency. In the field of laser
welding, the use of fibre lasers has been applied for precision welding of films, textiles and
high thickness moulded parts.
Diode Lasers
Diode laser (semiconductor laser) has a wavelength range from 808 to 980 nm. Diode laser
welding results in cleaner and stronger joints. The compact design, modular set-up and
high efficiency of diode lasers make them more convenient for industrial applications [25].
Diode lasers operate in the Near-InfraRed (NIR) spectral region, where thermoplastics
have low intrinsic absorption, implying that the use of this type of laser is highly dependent
on the presence of absorbent additives in the thermoplastics. Without the presence of
highly absorbing additives, the laser will penetrate the absorbent part without being
absorbed.
In the case of diode lasers, the weld spot size is usually not as small as Nd:YAG or
fibre lasers. Diode laser is usually sufficient for plastic welding as it is feasible to deliver
the energy using a combination of optical fibres and lens system [11]. However, diode
laser and its corresponding machinery is relatively affordable. Diode laser is suitable for
overlap joints such as lap-joint [26].
12
Laser transmission welding of bio-sourced thermoplastic composites
Laser power is one of the most critical parameters in the laser welding process [27]. Laser
power contributes to 58% of the welding process [28]. By increasing the laser power, the
heat input at the weld interface increases, and hence the amount of material melted would
be more in quantity with an enhanced weld strength. With an increase in laser power,
the depth of penetration and weld width increases [29]. However, laser power should be
combined with optimal laser speed to form an optimal bond.
Laser speed refers to the movement of the laser beam during LTW. Sufficient time has
to be given for melting and inter-diffusion of the two polymer surfaces. This can be
controlled by welding speed. The temperature distribution is also crucial for achieving a
good quality joint [30].
Focus position (or stand-off distance) is the distance between the surface of the materials
and the laser module. This parameter affects the weld seam width and temperature
distribution in the weld zone.
1.3 Bio-composites
Bio-based polymers are the polymers made from biological sources. Some of these poly-
mers are formed directly in the polymeric form within the producing organisms (ex.
micro-organisms, algae, or plants), while others are manufactured ex-vivo from bio-based
monomers. It is usually produced from fermentable sugar and vegetable oils. Since 1990,
advances in chemical processes and biotechnology have paved way to an increase in the
research conducted on bio-based polymers [31, 32].
Polylactic acid (PLA), polybutylene succinate (PBS), Polyhydroxyalkanoate (PHA),
and Polyethylene Furanoate (PEF) are some of the bio-based polymers that have re-
ceived much attention from many industrial players [31]. These new bio-based polymers
13
1.4. Interaction between the laser beam and the composite material
have the potential to replace petroleum-based polymers (PE, PP, PS, PTFE etc.) and
help solve some of the most urgent problems caused by the overuse of petroleum-based
polymers, such as water and soil pollution, deleterious influence to human health, and
over-dependence on petroleum [33].
Green reinforcement materials/ natural fibres are derived from plant and animal based
biomass. They are classified based on their origin and undergo separation and purification
treatment based on their origin. Some of the most common reinforcements are wood,
sisal, bamboo, abaca, and cotton. Agricultural biomass extracted from stems include,
flax, hemp, jute, kenaf, etc. [33, 34, 35, 36] The main objective to use natural fibres as
reinforcement in composites is to improve mechanical properties and produce lightweight
materials [37].
Bio-based composites, made from bio-derived polymers and natural fibre reinforce-
ments, are of growing interests in automotive industry. The European automotive indus-
try has been embracing the use of bio composites as vehicle components, due to their low
cost, low weight, ease in material availability, and most importantly mechanical charac-
teristics [38]. Moreover, the biodegradability, renewability, non-toxicity, and abundance
in nature make natural fibres an ideal reinforcing material for the composites for future
utilization in automotive industry [39].
The semi-transparency of a composite depends on the concentration and nature of
its constituents. Semi-transparent thermoplastic composites are composed of thousands
of fibres embedded in a polymer matrix. Semi-transparency thus appears as the general
behavior of matter.
Semi-transparent parts with opaque fibres such as flax fibres cause the occurrence of
three important phenomena during laser transmission welding: reflection, diffusion and
absorption, resulting in the attenuation of the laser irradiation as it passes the composite
in a given direction. The deposit of adequate energy at the weld interface is challenging to
14
Laser transmission welding of bio-sourced thermoplastic composites
control due to the heterogeneity of the semi-transparent part. Local fiber volume fraction,
composite architecture, refractive indices of the fibre and matrix, part thickness, are some
of the important properties of the semi-transparent part that affect its optical properties
[40].
Surface absorption
Semi–Transparent part
Short fibres
Laser beam
Weld interface diffusion
Reduction in
energy absorption z
Absorbent part
x y
Figure 1.4: Diffusion and absorption of laser energy in semi-transparent part due to the
presence of fibres.
15
1.5. Formation of a weld
or additives that are absorbent to laser. Fillers such as carbon black, carbon fibres, carbon
nanotube and MB black are very common in the industry [44, 45]. These nucleating agents
modify the material’s thermal, mechanical, and optical properties [46, 19].
Apart from energy loss due to absorption, reflective losses must also be addressed.
The reflectivity (ηi ) [47] at the interface of the two media having refractive indices n1 and
n2 is written as 2
n2 − n1
ηi = (1.2)
n2 + n1
The joining phenomenon of two polymers involves the interfacial reaction between the two
surfaces. Figure 1.5 shows the steps involved in formation of interfacial bond by fusion of
polymer chains.
Interfacial interdiffusion of polymer chains varies slightly between semi-crystalline and
amorphous thermoplastics. Semi-crystalline thermoplastics consists of crystalline and
amorphous regions [49, 50, 51]. They have a glass transition temperature (Tg ) in the
amorphous regions and crystalline melting temperature (Tm ) in the crystalline regions.
The crystalline melting point is higher than the Tg . These polymeric materials flow
above Tm , where no crystalline regions are present [52]. Crystal thickness and topological
constraints in the amorphous region define the melting of the crystalline component in
semi-crystalline polymers [53]. The chain mobility of the polymer increases above Tm or
Tg in the weld region. This allows the chains to diffuse across the joint interface and get
16
Laser transmission welding of bio-sourced thermoplastic composites
entangled along with the chains of another side of the interface, hence, forming a weld.
(iii) Wetting
(iv) Diffusion
(v) Randomisation
During healing, the diffusion stage is considered an essential step controlling the de-
velopment of mechanical properties. Healing stages are explained through functions such
as healing function, wetting distribution function and intrinsic healing function. These
functions relate weld strength and weld parameters to temperature, pressure, molecular
weight, time and processing conditions. A complete loss of memory of the interface occurs
during the randomisation stage. The inhomogeneity at the interface disappears at the
end of the diffusion stage. During diffusion and randomisation stages, the polymer chains
have more degrees of freedom of movement. Hence, the weld strength at the interface
appears during these last two stages[56].
17
1.6. Modelling of laser transmission welding
helps to effectively shorten the process time and efficiently obtain the ideal process pa-
rameters [20]. Many methods are available in literature that have modelled LTW process
to predict the developed weld strength in similar and dissimilar thermoplastics.
Bates et al. [57] performed finite element method (FEM) combined with a non-
linear model-fitting method to study the thermal degradation of Polycarbonate (PC) and
Polyamide 6 (PA6) containing carbon black during LTW. The influence of laser power on
polymer degradation has been studied, which implied a decline in weld strength.
Mayboudi et al. [58] developed a two-dimensional FEM model to investigate the LTW
of PA6. The depth of the heat affected zone and temperature distribution in the weld
zone were predicted, The method was extended to a three-dimensional (3D) model to
calculate the spatial state changes of materials during the welding process. A MATLAB
code was developed to solve the 3D transient heat equation as a function of laser beam
dimensions, laser beam power and absorption coefficient of the transparent and absorbent
parts [59].
Geiger et al. [60] investigated the effect of absorption coefficient of Polypropylene
(PP) on the temperature field distribution and volume of the heat affected zone using
FEM. It was concluded that the geometry of molten pool was negligibly affected by the
absorption coefficient change with the temperature field.
Labeas et al. [61] developed a 3D thermo-mechanical FE model for LTW process to
analyze the influence of process parameters on the real-time welding process by combining
multi-physics fields.
Aden et al. [62] modelled LTW process to simulate the temperature field at the weld
interface. Interface temperature distribution sourced by Gaussian and M-shaped laser
intensity distributions was compared on COMSOL Multiphysics.
Asseko et al. [63] developed an FEM thermal model in uni-directional glass fibre
reinforced composites. The laser scattering and absorption phenomena were coupled to
predict the temperature distribution at the heat affected zone. The model was devel-
oped to determine the ray path during propagation through heterogeneous medium and
calculate the absorption coefficient [64].
More approaches to optimize the welding process are found in the literature, apart
from FEM. Acherjee et al. [27] used response surface methodology (RSM) to predict the
18
Laser transmission welding of bio-sourced thermoplastic composites
welding strength and seam width in the LTW of acrylic sheets. The same method was
applied to dissimilar plastics welding, Polycarbonate to ABS [65]. RSM has been used
to develop empirical models to relate welding parameters to weld strength by performing
many experiments with varied parameters. The researchers also studied the distribution
of temperature along the depth of materials in the LTW of PA6 using thermal imaging
and theoretical calculations [66].
Artificial neural network is a developing tool which has been applied to predict weld
strength and weld seam width of laser welded acrylic sheets [67]. ANN is proved to provide
a high precision in process optimisation [68]. However, RSM and ANN methods require a
history and a database of weld procedure performed with all possible variations of process
parameters and materials.
Strength development at a weld interface can be analysed with mechanical testing. Me-
chanical tests help to evaluate healing relations for polymer interfaces [21]. Generally,
mode I and mode II fracture tests are performed.
The mechanical energy G required to separate two welded parts is a function of time
t, temperature T , contact pressure P and molecular weight of the polymer chains MW :
G = W (T, P, t, MW ) (1.3)
r
σ G
= (1.4)
σmax Gmax
where σmax is the ultimate strength of the bond and Gmax is the corresponding maximum
energy release rate.
19
1.7. Determination of weld strength with mechanical tests
(a)
(b)
(c)
(d) A
Figure 1.6: Various fracture tests to determine the interfacial bond strength [69, 70, 71,
72]
20
Laser transmission welding of bio-sourced thermoplastic composites
Mechanical tests, especially shear strength tests, are widely used to measure the me-
chanical performance of the weld joints. The two most used geometries in laser transmis-
sion welding samples are lap-joint and T-joint.
Yan et al. [69] applied wedge test (Figure 1.6 (a)) to characterize the bond strength.
It is a mode I fracture test with fracture energy calculated as,
3Et3 ∆2
GIC = (1.5)
16a4
referring to Figure 1.6 (a) where E is the Young’s modulus of the materials, t is the
thickness of the plates, ∆ is the length of the opening, and a is the crack length.
Another mode I fracture test is the double cantilever beam test [70]. This test requires
additional support parts to apply load. Figure 1.6 (c) is a schematic of this test. The
fracture energy is determined as
3P δ
GIC = (1.6)
2B(a + ∆)
where, δ is the displacement of the specimen when a load P is applied, and B is the
specimen width. ∆ is the crack length parameter, calculated from linear regression of the
cube root of the compliance, against the crack length a [70].
End notched flexure (ENF) test requires generation of a weld line at the weld interface
along the length of the materials, leaving approximately one-third of the length without
a weld. Referring to Figure 1.6 (b), the mode II critical strain release rate with direct
beam theory [71] can be written as:
9a2 P δ
GIIC = (1.7)
2B(2L3 + 3a3 )
Fmax
σw = (1.8)
A
21
1.8. Objectives
where, A is the area of the bonded region and Fmax is the maximum tensile load.
1.8 Objectives
22
Laser transmission welding of bio-sourced thermoplastic composites
23
Chapter 2
25
This work presents a numerical study of the effect of fibre orientation in fibre-reinforced
composites during the laser transmission welding simulation.
The focus of this chapter is the light scattering that leads to the change in shape and
intensity of the laser beam at the weld interface, which further influences the temperature
field that forms a weld [78, 63].
A three-dimensional (3D) numerical structure is generated to take into account the
microstructure of a composite material reinforced with short fibres. The information of fi-
bre volume fraction, fibre length distribution and fibre orientation distribution are needed
to generate the numerical microstructure. This information is extracted by numerically
simulating the manufacturing process. In the current context, the manufacturing pro-
cess used is injection moulding. The fibre orientation distribution function can then be
calculated from fibre orientation tensor.
The mechanical properties of composite materials are a function of fibre orientation.
Various works [76, 77] have shown the ability to predict the influence of the orientation
distribution of fibres on the mechanical behaviour of materials. Manufacturing process
conditions and matrix flow patterns can be linked to fibre orientation distribution in SFRT
composites. The reflection and refraction of light transmitted through SFRT composites
are studied to predict the influence of fibre orientation on laser transmission welding of
SFRT.
An algorithm is implemented to trace rays propagating through the composite mate-
rial. This algorithm uses the optical properties of the composite material to simulate laser
beam reflection and refraction in a complex structure. The laser beam scattering at the
weld interface is simulated in 3D geometries with various fibre orientation distributions.
The effect of fibre orientation on the light scattering phenomenon of laser is compared for
different cases of fibre orientation. The simulation results are validated experimentally.
During the injection moulding, flow conditions determine the orientation of short fibres
[79]. A common feature of fibre orientation seen in injection moulded parts is the shell-
core distribution. The orientation of fibres at the part shell is parallel to the flow, whereas
it is perpendicular at the core. Highly aligned fibre orientation is also observed at the
inlet and exit of a mould. The direction of fibres within the material strongly impacts
the mechanical properties [80, 76, 81, 82]. The influence of the shell-core distribution on
26
Laser transmission through fibre reinforced composites
27
2.1. Construction of 3D numerical geometry
(a) (b)
Figure 2.1: Grid used in algorithm showing (a) inclusion interaction (b) inclusion centres
and periodicity [88].
coordinates. The placement of the inclusions in the geometry is done with level sets. This
approach (as seen in Figure 2.1(a)) avoids overlapping of the inclusions in the 3D grid as
in Poisson’s process. This cell is also generated with periodicity (Figure 2.1(b)).
The model creates polyhedrons to represent the fibres in a 3D grid. The algorithm uses
the parametrization of an ellipsoid to define the inclusions. Several geometric shapes can
be included in the geometry, such as ellipsoids cylinders, cuboids or regular polyhedrons.
As the algorithm allows regular polyhedrons to represent fibre inclusions in the numerical
geometry, it permits the representation of complex shapes of natural fibres in the geometry
with simple modifications. To simplify the simulations in this paper, the fibre shape is
selected to be cylinders as are glass fibres[89].
28
Laser transmission through fibre reinforced composites
The information on the fibre length distribution in a reference was used in this work. Based
on injection moulding simulations of short fibre composites on Autodesk MOLDFLOW, a
fibre length and radius analysis was performed by Soete et al. [90]. The average length of
the fibres was 500µm and the radius was 100µm. This data is used to define a probability
distribution function to assign the length of the fibres in the grid.
During the numerical simulations, a laser beam is discretized into many rays to trace
their paths. Every ray that travels through the numerical geometry should be obstructed
at least once in its path to avoid a ray concentration effect. From the viewpoint of
the laser source, the generated geometry can be visualised using image processing. In
a monochrome image (Figure 2.2), two regions can be distinguished: the black area
represents the fibres, and the white area represents the region where the laser beam
can travel through the numerical geometry without any obstructions in its path. This
criterion intends to construct the RVE while minimising the white areas, to obtain the
direction of ray path deviation by imposing at least one interaction between fibres and
laser. White lines shown in Figure 2.2 are the outlines of fibres which are represented for
the sake of illustration.
The numerical geometry construction converges to about 99% obstruction (black region)
and can not reach 100% without significantly increasing the fibre volume fraction. To
reduce the computation cost, convergence is preferred over achieving a full obstruction
to ray paths. This also risks allowing a few rays to travel uninterrupted to the weld
interface and these rays generate a local heat concentration at the weld interface. This
circumstance is also expected, however, during light diffusion with complete obstruction.
The fibre volume fraction converges at 29% maintaining the convergence of the black
region.
29
2.1. Construction of 3D numerical geometry
Empty volume
The prediction of fibre orientation state using orientation distribution function which is
a kind of a probability density function, by F. Folgar and C. Tucker [76] and its exten-
sion into use of Fibre Orientation Tensor (FOT) by S. Advani and C. Tucker [87] have
been a common approach to investigate the influence of fibre orientation on thermal and
mechanical properties of composite materials [91].
The orientation of the fibre inclusions in the numerical geometries is defined using
rotation matrices with respect to axes 1 and 3. These angles of rotation are ϕ and θ, and
are obtained from fibre orientation tensors. FOTs are simplified descriptions of orientation
distribution functions defined by the orientation of a unit orientation vector p with respect
to coordinate axes as shown in Figure 2.3. A second-order FOT is written as
⟨p p ⟩ ⟨p1 p2 ⟩ ⟨p1 p3 ⟩
1 1
a = ⟨p2 p1 ⟩ ⟨p2 p2 ⟩ ⟨p2 p3 ⟩ (2.1)
⟨p3 p1 ⟩ ⟨p3 p2 ⟩ ⟨p3 p3 ⟩
where, the angle brackets denote a configuration average, i.e., an average over the range
of p, weighted by the probability distribution function for orientation ψ(p).
30
Laser transmission through fibre reinforced composites
I
aijkl = pi pj pk pl ψ(p) dp (2.3)
p3 = cos θ (2.4c)
31
2.1. Construction of 3D numerical geometry
Hence, a second-order FOT consisting of 9 elements, has only five independent elements
that contain most of the quantitative information about fibre orientation.
To concisely describe the orientation distribution function of the fibre inclusions, fibre
orientation tensors are formed. Fibre orientation tensors are defined by forming dyadic
products of vector p and integrating the product of these tensors with the distribution
function over all possible directions.
The orientation vector p can be used to describe the orientation distribution function,
ψ with deviatoric versions (bij , bijkl ) and tensor basis functions (fij , fijkl ).
1 15 315
ψ(p) = + bij fij (p) + bijkl fijkl (p) + ... (2.6)
4π 8π 32π
Fibre orientation tensors hold information about the orientation which can then be
described by the orientation distribution function (ODF). The set of all possible directions
of p corresponds to a unit sphere. A second-order tensor is a set of nine elements out of
which only six are unique due to symmetry[92].
3 3
1 1
2 2
(a) (b)
Figure 2.4: Orientation distribution of fibre shown in a unit spheres, generated from fibre
orientation tensor with components a11 = 1 and a22 = a33 = 0. (a) second-order ODF (b)
fourth order ODF.
32
Laser transmission through fibre reinforced composites
Figure 2.4 shows that increasing the order of the FOT increases the accuracy. Hybrid
closure approximation (aHYB
ijkl ) [94] is widely used to extend the second order orientation
tensor to the fourth order in numerical simulations of injection moulding process [95, 82].
This closure approximation is a sum of two other closure approximations: linear (aLIN
ijkl )
1
aLIN
ijkl = (aij δkl + aik δjl + ail δjk + akl δij + ajl δik + ajk δil )
7 (2.7)
1
− (δij δkl + δik δjl + δil δjk )
35
aQUA
ijkl = aij akl (2.8)
QUA
aHYB LIN
ijkl = (1 − f )aijkl + f aijkl (2.9)
where, f = 1 − 27 det(aij ).
It has been observed that when using hybrid closure approximation, around 1% of the
orientation angles θ and ϕ are out of the expected range of orientation. These samples
can be seen in Figure 2.4 (b) as a vertical orthodrome. These angles are not eliminated
during geometry generation, in this work. Von Neumann rejection algorithm [96] is used
to sample θ and ϕ from the fourth order ODF. A composite microstructure is generated
from fibre orientation tensors using the algorithm implemented in MATLAB (See Figure
2.6 ).
The algorithm is verified by sampling many pairs of orientation angles and construct-
ing a composite geometry by randomly selecting pairs from the samples. Subsequently,
orientation angle information is collected from the fibre inclusions in the geometry and an
FOT is regenerated from it (Figure 2.6). Comparing the resulting fibre orientation tensor
with the initial input fibre orientation tensor, it can be confirmed that more than 99%
accuracy is achieved and the method is verified.
33
2.1. Construction of 3D numerical geometry
Yes
Vf ≥20% ? NO
Yes
W=0? NO
White area
Yes
Save inclusion center
y
coordinates, dimensions,
angles of orientation
x
Create microstructure in
COMSOL, build mesh,
save mesh file
END
Figure 2.5: Algorithm to generate 3D numerical geometries to perform ray tracing simu-
lation
34
Laser transmission through fibre reinforced composites
𝜙 𝜃
Figure 2.6: Method verification to use fibre orientation tensors to define orientation dis-
tribution function.
35
2.2. Ray tracing method
The Ray tracing method is widely used in the context of computer-generated imagery,
which follows the path of a photon emitted from a source to locate the illuminated areas
in its environment. The method is based on mathematical algorithms of the intersection
of surfaces (illuminated objects) with straight lines in space (photon paths). Ray tracing
was first applied in 1883 by Christiansen [97], in the context of radiative calculations. The
ray tracing method consists of simulating the real path of the rays emitted by the laser
beam and their interaction with the object studied while taking into account reflections
and refractions.
The ray tracing method is based on geometrical optics. In a complex medium such
as a composite structure, the propagation of laser rays can be simulated using the ray
tracing method which is based on the Snell-Descartes law [78].
where, ni and θi (i = 1, 2) are the refractive indices of the medium i and the angle
formed by the ray and normal at the interface (See Figure 2.7). The circular laser beam
is discretized into rays which are distributed in a Gaussian or square form.
Incident ray
𝜃1
Medium of index 𝑛1
Medium of index 𝑛2
Refracted ray
𝜃2
36
Laser transmission through fibre reinforced composites
The previously constructed 3D geometry is then meshed with the finite element
method. In this heterogeneous medium, the change in path and direction of all laser
rays is calculated at each instance of change in the medium at a micro-scale. Each ray
interacts with every element in its individual path and changes direction based on the
change in the refractive index if the medium is alternated.
C 𝑋
Tetrahedral element
When there is a change in the medium, the tetrahedral element at the intersection
plays a role to change the ray direction. The ray enters through one face and its exit face
is calculated based on the ratio of the refractive index of the media involved. To find this
exit face, the ray tracing simulation tests all the faces of the tetrahedron element into
which the ray has entered. For the triangular face (ABC), in an orthonormal frame of
→
−
reference with any centre O, and a ray entering from point D and direction vector V ,
there is intersection when the following condition is reached (See Figure 2.8).
37
2.3. Numerical procedure to determine optical diffusion
where u, v, t are the components of the vector X which are calculated by Equation 2.12.
−−→ −−→ →
−
The new starting point D is calculated as OD → OD + t V . The approach is recalled
→
−
and the direction vector V is recalculated whenever the ray crosses an element face which
connects two elements of different refractive indices. In fact, the deviation of rays at the
micro-scale is the laser diffusion at a macro-scale [64].
In previous studies [98, 6, 17], the ray tracing method was used to perform simulations
of the ray propagation in the semi-transparent part. This method allows for calculating the
energy that reaches the weld interface and its location. Even though it is possible to couple
the ray tracing method with Beer-Lambert absorption law to determine laser energy lost
during the propagation, the absorption in the semi-transparent part is considered to be
negligent, in this section.
38
Laser transmission through fibre reinforced composites
approaches the representation of case 1 only with very high-order tensors. Referring to
Figure 2.3, these fibre inclusions are perfectly parallel to axis 1 with orientation angles
θ = π/2 and ϕ = 0. Figure 2.10 visualises the orientation of the fibre inclusions in each
case of the numerical geometries.
3 3 3
1 1 1
2 2 2
Case 1 Case 2 Case 3
The mean intensity of rays that reach the weld interface is obtained by gradually
increasing the projected number of rays. Convergence is reached at 200000 rays (See
Figure 2.11).
39
2.3. Numerical procedure to determine optical diffusion
40
Laser transmission through fibre reinforced composites
The implemented ray tracing algorithm reduces the computation cost by using peri-
odic geometries. The numerical geometries are repeated multiple times in all directions
summing the thickness of the numerical geometry up to 3 mm and the interface surface to
a 6 mm x 6 mm plane. The non-focused laser beam diameter is 1 mm and the intensity of
the rays is square distributed. The ratio of refractive index of fibre and matrix is varied
with 5% as proposed by Cosson et al. [64] as nm /nf = 1.44/1.55.
41
2.5. Results and discussion
isotropic orientation distribution along the thickness of 3mm (Figure 2.12 (a)). On this
square area, each component of the fibre orientation tensor is averaged at each layer, i.e.
at each thickness position. Figure 2.12 (b) represents average fibre orientation tensor
components a11 , a22 and a33 values along the thickness of the plate in 13 layers.
(a) (b)
Figure 2.12: (a)Location on injection moulded short-glass fibre reinforced PLA composite
for fibre orientation distribution information, (b)Variation of homogenised fibre orienta-
tion distribution in thickness of sample plate.
The optical properties of the thermoplastics in the two plates can be easily obtained
because both PC and PLA are completely transparent and are commonly used as trans-
parent parts during laser transmission welding [99, 100]. The difference in the refractive
index of the fibre and matrix is close to 5% similar to the optical properties input in
the numerical simulations. A laser module of 5mW power with a circular beam of 1mm
diameter is considered for experimental observations.
The result of the ray tracing simulation is the number intensity of the rays and their
location after the rays have travelled through the numerical geometry and reached the
weld interface. This was demonstrated by projecting 200000 rays of a laser beam of 1 mm
diameter into two numerical geometries of 3mm thickness, one with no fibre inclusions
42
Laser transmission through fibre reinforced composites
1 0 0
(indicating a pure material) and the other with the fibre orientation tensor
0 0 0. A
0 0 0
map of the diffusion of laser rays which have propagated through the two geometries is
generated (Figure 2.13).
When there are no fibres in the numerical geometry, the laser rays travel along straight
paths showing no signs of scattering. On the contrary, when the rays travel in the com-
posite, the ray paths are disturbed causing a deviation from their original paths. More
importantly, it has been observed that the diffusion occurs in the direction perpendicular
to the fibre inclusion orientation. Additionally, not all rays reach the target end, as they
may have reflected back or refracted out from other faces of the numerical geometry. It is
also possible that some of the rays circulated within the numerical geometry in an infinite
loop.
Laser rays
Initial ray dimension
Number of rays
Transparent
part
z
Weld interface
y x
y
Sample length x
Figure 2.14 compares the diffusion map of 200000 rays for different orientation states.
On average, 160000 rays successfully reach the weld interface in all 3 cases. The other
rays have either exited the numerical geometry through the other faces or have refracted
inside the numerical geometry in an infinite loop (which implies that the energy of these
rays is completely absorbed in the numerical geometry; this case is not considered in this
work). Additionally, the influence of 99% obstruction of rays is seen in some parts of
Figure 2.14 in red areas. There is a high concentration of laser rays in these areas which
43
2.5. Results and discussion
Number of rays
implies that 100% obstructed region is not achieved. The 100% obstruction region results
in a high fibre volume fraction and corresponds to high computation costs. Analysing the
distribution of rays that reach the exit face of the geometry, it is quite clear in Figure
2.15, that the laser rays have scattered significantly. Especially in Case 1 (perfectly
unidirectional orientation, Figure 2.15 (a)), the rays diffuse perpendicular to the fibre
orientation. At the location of the fibre interaction, the rays have only two paths for
refraction: either the perpendicular or the longitudinal to the fibre direction.
Number of rays
Pixel
1 0 0 1/3 0 0
𝜋
Perfectly unidirectional 𝜃 = &𝜙 =0 Fiber orientation tensor = 0 0 0 Fiber orientation tensor = 0 1/3 0
2
0 0 0 0 0 1/3
Rays reaching pixels in central y-axis
Rays reaching pixels in central x-axis
Figure 2.15: Rays at the central lines of the interface in x and y directions.
It is clear that the results for Case 1 and Case 2 should be very similar because the
fibre orientation is similar for these two cases. It should be noted that the cylinder shape
of the fibres has made the results explicit.
44
Laser transmission through fibre reinforced composites
The influence of the low-order tensors is quite evident when comparing the diffusion for
Cases 1 and 2. Even if the simulation results are expected to be similar for these two cases,
there is a notable difference due to the incomplete fibre orientation information contained
in the fourth-order tensor. The slight misalignment of the fibre inclusions has caused the
diffusion in the direction of the fibre orientation. It is also possible that the laser rays
reaching the ends of the inclusions are free to diffuse in various directions. One such
encounter in a ray path could disturb the expected pattern of diffusion at the interface.
To verify this, a numerical structure with longer fibre inclusions is constructed and the
diffusion pattern is compared (Figure 2.16). The laser ray diffusion is more compacted in
the centre with less scatter in the direction of the fibre inclusions. Hence, the length of
the fibres influences the diffusion of rays.
More disruptions of the ray path lead to greater diffusion of rays. It may be concluded
that the ray has propagated in the semi-transparent part by reflecting and refracting,
resulting in a significant attenuation of energy before reaching the weld interface. This
attenuation can be strongly dependent on the fibre volume fraction in the semi-transparent
part. A high number of rays reaching a single location of the interface results in high
absorption of energy locally which can lead to the degradation of the polymer. With a
45
2.6. Conclusion
controlled initial beam intensity, it is possible to associate the fibre morphology such as
fibre volume fraction, fibre shape, fibre length and fibre orientation to the optimal weld
parameters such as laser beam intensity, laser diameter and laser beam speed. An increase
in fibre volume fraction decreases the possibility for a laser beam to propagate through
the material without scattering [101].
The simulations for Cases 1 and 3 were validated experimentally by projecting a laser
beam through composites with similar fibre orientation. Figure 2.17 shows these two
cases with glass fibres. The initial shape of the beam was observed by projecting the
laser through a thin paper and a picture was captured from the other end with a digital
camera (Figure 2.17 (a)). Replacing the paper with unidirectional and randomly oriented
glass fibre reinforced composite plates, the laser beam shapes as shown in Figure 2.17 (b)
and (c) are observed respectively. Evidently, unidirectional fibre composite diffuses the
laser beam in the direction perpendicular to the fibre direction, which corresponds to the
ray tracing simulations result. It is also seen that the random fibre orientation of fibres
diffuses the laser in all directions which are also similar to the simulation result.
It should also be noted that the ratio of refractive index greatly influences the quantity
of diffusion. The dimensions of the diffused laser spot can be altered by changing the
optical properties of the fibres. Replacing the glass fibres with more opaque fibres such as
flax, the diffusion can be expected to be greater [64]. The energy absorption at the weld
interface can be expected to be lower, even if the energy absorption in the semi-transparent
part is neglected.
2.6 Conclusion
The influence of the fibre orientation on the laser beam diffusion was numerically modelled.
The distribution of the laser energy that reaches the weld interface can be used for FEM
simulations to compute the temperature field at the weld interface, which is helpful for the
optimisation of laser transmission welding of composites. A numerical model to generate
microstructures representing composite materials considering fibre volume fraction, fibre
shape, fibre length and fibre orientation was developed. Considering the influence of
fibre orientation on both the mechanical properties and the laser welding quality, it is
46
Laser transmission through fibre reinforced composites
Figure 2.17: Laser beam shape after propagating through (a) thin paper, (b) unidirec-
tional glass fibre reinforced composite, and (c) randomly oriented glass fibre reinforced
composite.
significant to take into account the real fibre orientation in the simulation of the laser
welding process.
47
Chapter 3
3.1 Introduction
During laser transmission welding, the extent of melted polymer volume at the weld
interface depends on the quantity of energy delivered between the materials to be joined.
Insufficient energy will not melt the material at the interface, whereas, a high amount
of energy may break cohesion between fibres and matrix or cause deformations [41]. As
seen in the previous chapter, the transmittance of the laser energy is influenced by the
obstructions in the laser path. The problem is therefore not confined only to welding
process parameters such as the laser wavelength and the laser power. Optimization of
the process parameters, depending on the optical and thermal properties of the involved
materials is required to avoid poor weld quality [61]. This can be done by linking the
weld interface temperature field to the welding process parameters and relevant material
properties.
A 3D transient thermal model of LTW is considered to obtain the temperature field
between the assembly of substrates by solving the energy balance equation:
∂T →
−
ρCp = −∇.(−k ∇T ) + Q (3.1)
∂t
where ρ is the material density, Cp is the specific heat, T is the temperature, t is the time,
k is the thermal conductivity, Q is the heat source and ∇ is the gradient operator.
To model the heating phase, firstly, a comparison between the intensity of the initial
49
3.1. Introduction
laser beam and that of the rays that have reached the weld interface must be carried
out. This allows a better understanding of how specific materials behave during laser
transmission welding. For semi-transparent composite materials, a substantial portion
of the light transmitted through the material may be deflected from the incident beam
direction, reducing the energy density at the interface. Furthermore, the effective laser
beam path length is now greater than the actual part thickness, thereby increasing the
part absorbance and decreasing laser transmission. It is therefore critical to assess the
strength of the reinforcements in the composite material [102].
Generally, laser energy absorption is considered only at the weld interface and in the
absorbent part [42, 43, 63]. These analytical approaches take into account the Beer-
Lambert law to calculate the intensity profile during laser propagation through the semi-
transparent part.
I = I0 e−Kd (3.2)
where I0 is the initial intensity of the laser beam, Kh is the absorption coefficient of the
material, d is the distance travelled by the laser beam in the material and I is the laser
the final laser intensity at the weld interface. This relationship indicates that the laser
beam scattering attenuates laser intensity exponentially [44]. This mathematical function
is not applied to intensity magnitude but to the total laser power and the parameter of
the exponential term is the absorption coefficient.
Beer-Lambert law has been commonly applied to the absorbent part only, while the
absorption in the assumed transparent part is neglected [73]. This is considered an ideal
situation during laser transmission welding [40]. The heat source in these cases of welding
is described with the following equation:
0
for z < dt
Q(x, y, z) = (3.3)
I(x, y) × K × e−K×dt
for z > dt
where dt is the thickness of the (semi-) transparent part and z is the coordinate axis
parallel to the direction of the laser beam. The absorption coefficient is hence considered
infinite for the absorbent part. Generally, the intensity distribution at the weld interface
is assumed to retain the initial shape of the laser source beam. This assumption can no
50
Heat transfer analysis
longer be applied in this work due to the heterogeneity of the semi-transparent composite
material.
By considering the calculations of distances travelled by the rays in the semi-transparent
part, and then applying a volumetric heat generation expression according to Beer-
Lambert law, one can determine the source term in the heat equation, Eq 3.1. The
model takes into account both maximum attenuation and the standard deviation of the
expansion of the energy distribution into the semi-transparent part. This method requires
homogenized coefficients of composites which have to be determined beforehand.
The thickness of the semi-transparent composite materials is one of the factors that
affect the absorption of energy in the part. As the thickness of the semi-transparent part
increases, the attenuation of initial intensity caused by the diffusion of the laser rays also
increases [103]. Though absorption is a function of laser wavelength [38], the laser module
is not changed in this work.
51
3.3. Experimental procedure
surface nor within the semi-transparent part may not be sufficient to melt the polymer
at the weld interface causing a bad weld or no weld at all.
Laser beam
Surface absorption
Transparent part
Absorption in
transparent part
Absorbent part
Absorption at
interface
Hence it is crucial to understand the aspects that cause the surface fusion marks and
the weld parameters that can reduce the surface absorption. Firstly, to apply the Beer-
Lambert law, the total length of the ray path much be taken instead of the part thickness
[64]. The term Kh is the homogenised absorption coefficient of the semi-transparent part.
The ray tracing simulation is carried out with the same physics as seen in the previous
chapter, but with a few modifications. In addition to following the path of each laser ray,
the simulation records the total distance travelled by the rays at a micro-scale.
52
Heat transfer analysis
Digital camera
Composite plate
(PLA+10% Flax)
White screen
of focus position on the absorption of laser radiation and find an optimal focus position
to maximise absorption at the weld interface instead.
3.3.1 Materials
To manufacture the composite plates, flax fibres with polylactic acid (PLA) are chosen
to emphasize on bio-sourced materials. For the experiments, a composite plate with PLA
and short flax fibres is taken as the semi-transparent part and a pure PLA plate with 1%
MB Black is taken as the absorbent part. Both the plates are 3 mm thick with dimensions
of 80 mm x 80 mm. The plates are cut into multiple samples for welding, with dimensions
18 mm x 80 mm.
The fibre volume fraction (vf ) of the semi-transparent part is 10%. The plates are
manufactured by injection moulding. The overall fibre orientation across the plate thick-
ness can be considered isotropic [105]. However, the fibre distribution in the plate is
heavily influenced by the flow conditions of the materials during injection moulding. Due
to this, weld line properties are predicted to differ in different regions of the plate.
53
3.3. Experimental procedure
Optical transparency
Figure 3.3: Fibre volume fraction distribution in the semi-transparent composite plate
The composite plate with short fibres has a non-uniform distribution of fibres. This re-
stricts the regions that can be welded for comparative studies. An experimental procedure
is carried out to map the fibre distribution in the composite plates and choose areas on
the plate with similar fibre volume fractions.
30 composite plates with a thickness of 2mm and 10% fibre volume fraction are pho-
tographed with a digital camera. A white screen with uniform distribution of brightness
is placed behind the plate (Figure 3.2). The experimental setup is arranged in darkness to
eliminate the influence of external light on the image. The parameters of the digital cam-
era such as exposure time aperture, white balance, ISO and focus are kept constant [106].
The images are captured in grey scale and image processing is carried out on MATLAB.
Figure 3.3 shows the distribution of fibres averaged in 30 composite plates. It can be
noted that there is a very low fraction of fibres close to the inlet of the injection mould
and there is an accumulation of fibres at the other end. These two regions may not be
suitable for welding due to a drastic difference in fibre volume fraction when compared to
54
Heat transfer analysis
Laser optic
Semi-transparent part
Absorbent part
3D printed support part
Weld bench
Clamping system
A simple variant of a fixed clamping system [107] is seen in the welding machine at IMTNE.
The materials are placed between a transparent cover and a work bench which is displaced
by a pneumatic actuator (Figure 3.5). For welding of high temperature polymers, quartz
55
3.3. Experimental procedure
glass is used as the transparent cover. Apart from being able to handle load from the
clamping pressure, the transparency of the transparent cover is also influential. Figure
3.6 shows that about 95% of laser energy is transmitted through the quartz glass cover
between the wavelength range of 300nm and 2000nm. The diode laser used in the current
machine is well within this wavelength range.
Transparent cover Laser
radiation
Pneumatic
actuator
Figure 3.5: Fixed clamping mechanism during LTW process [108, 107]
Transmission [%]
Wavelength [nm]
Figure 3.6: Transmission of laser beam through quartz glass with respect to laser wave-
length [109].
Support part
During welding process, the weld bench with the specimens moves towards the welding
area under the transparent cover. During this displacement, the specimens are at risk of
dislocation. A 3D printed support part is designed and manufactured to fit on the weld
bench with dimensions based on the specimen size.
56
Heat transfer analysis
Welding procedure
The semi-transparent part is placed above the absorbent part on the weld bench and fixed
in a location with the support part. The process parameters are input on a screen of the
welding machine. Figure 3.7 shows the options available to insert welding parameters.
Firstly, the screen shows “Laser welding” slide where four icons can be seen. “Coordi-
nates” icon allows to input 2D coordinates of the start and end points of the expected weld
line. These values can be found using rulers attached to the weld bench. Additionally,
the laser movement speed is input here.
“Parameters” icon leads to options where the laser power is chosen between 0 and 46
W. Other options include laser preheating time, holding time after welding and number
of passes of the welding.
The “motor” icon calibrates the position of the laser module before starting the welding
process with the icon “start cycle”.
MOTOR
𝑌2 Point 2
CALIBRATE
𝑌1 Point 1
RETURN
𝑋1 & 𝑋2
57
3.3. Experimental procedure
During the welding of a semi-transparent part with a 10% fibre volume fraction, a fusion
mark was observed in many cases. The surface fusion mark is caused by the absorption
of laser energy at the surface of the semi-transparent part. Weld parameters such as laser
speed, power and focus position were iterated to determine the influence of each of these
parameters on the formation of the fusion mark. It was concluded that focus position
plays a major role in forming the burn mark.
72mm
70mm
Focus position 72 mm
70 mm
65mm 65 mm
The focus position is the distance from the laser beam module to the surface of the
transparent part. It ranges from 72 mm to 52 mm in the welding machine at IMT Nord
Europe. The focal angle is calculated experimentally by measuring the radius of the
projected laser beam at every focus position and outlining the triangular structure (2D)
formed by all of these projections. The focal angle θ is 8.1712°.
The next step is to find the focus position at which the burn mark is formed and
another position at which there is no burning mark. Three focus positions are chosen to
weld and the results are seen in Figure 3.8. It is seen that with the farthest focus position,
a burn mark is formed at the surface of the semi-transparent part. As a slightly lower
length at 70 mm, the burn mark appears to be slightly diminishing non-uniformly along
the length of the fusion mark. Finally, when the laser module is at a distance of 65 mm
from the material surface, there appears to be no surface fusion mark.
58
Heat transfer analysis
Focal angle
𝜃 = 8.1712 Focus position=
72mm to 52mm
Materials to weld
The numerical simulation of energy exiting the semi-transparent part is quantified with
the Monte Carlo method which is the basis of the ray tracing method. The two media in
the path of the laser rays differ extremely in optical and thermal properties, resulting in
scattering and absorption. The intensity distribution of the laser beam within the joining
plane is expanded and the laser power usable for the welding process gets reduced [110].
The effect of focus position on the surface absorption of laser radiation is investigated.
A 3D numerical geometry is generated that represents the optical and thermal properties
of the semi-transparent part. A heat transfer simulation is carried out to determine the
temperature distribution in both the semi-transparent and absorbent parts.
59
3.4. Numerical method
The geometry generation in the case of this simulation is considered with a few con-
straints. Firstly, the shape of the inclusion is taken as spheres. This is to eliminate the
effect of fibre orientation on diffusion, which clearly is influential as seen in the previous
chapter. Also, the volume fraction is limited to 10% without any consideration of com-
plete obstruction in the beam path. The only changing variable in the current study is
the focus position.
The geometry is repeated multiple times in each direction with the aim to achieve
a part thickness of 3mm and a plate size larger than the cross-section area of the laser
beam.
120
120
3 mm 20
Cubic geometry with spherical inclusions and periodicity. Discretisation Geometry multiplied to obtain 3 mm thickness
As the heat equation is solved in a transient state, the thermal properties such as
material density, thermal conductivity and heat capacity act on the temperature field
during welding. The thermal properties of PLA and flax at ambient temperature are
provided in the table.
60
Heat transfer analysis
Table 3.1: Thermal properties of PLA and flax fibres at ambient temperature
The specific heat capacity (Cp ) at constant pressure describes the amount of heat energy
required to produce a unit temperature change in a unit mass. The temperature depen-
dence of the specific heat capacity of PLA is taken into account. The specific heat of flax
is assumed to remain constant at all temperatures.
In this work, the heat capacity of the composite material is calculated with the exact
rule of mixtures.
Cp,c (T ) = vf Cp,f + vm Cp,m (T ) (3.4)
where, Cp,c , Cp,f and Cp,m are the specific heat capacities of composite, fibres and matrix
respectively.
ρc = vf ρf + vm ρm (3.5)
61
3.4. Numerical method
Temperature [℃]
Thermal conductivity
For most polymers, the primary mechanism of heat conduction is through phonons [115]
The two basic models representing the upper bound and the lower bound for thermal
conductivity of composites (kc ) are the rule of mixture (parallel model) and the series
model, respectively [116]. In the parallel model, each phase is assumed to contribute
independently to the overall conductivity, proportionally to its volume fraction:
kc = vf kf + vm km (3.6)
On the other hand, the series model assumes no contact between particles and thus
the contribution of fibres is confined to the region of the matrix embedding the particle.
The conductivity of fibre composites according to the series model is predicted as
1
kc = (3.7)
(vm /km ) + (vf /kf )
The thermal conductivity (k) describes the relationship between the heat flux vector
(q) and the temperature gradient (∇T ) with Fourier’s law of heat conduction.
q = −k∇T (3.8)
62
Heat transfer analysis
However, in the current geometry, the spherical shape of the fibre inclusions has influ-
enced the geometry equally in all directions. The effective thermal conductivity (kef f ) in
the directions x, y and z is calculated to be 0.18 W/m.K.
63
3.4. Numerical method
Absorption coefficient
lm lf
Kh = Km + Kf (3.10)
lm + lf lm + lf
Hence the distance travelled by a ray in the matrix volume (lm ) and fibre volume (lf )
should be calculated at a micro-scale during ray-tracing simulations (Eq. 3.10).
For the high number of rays projected into the heterogeneous geometry with numerous
fibre-matrix interactions, the calculations of distance travelled by all the rays consume
high computation costs. Cosson et al. [64] observed that the ratio of the average distance
travelled by a laser ray in the fibre to the total average distance can be equated to the
lf
fibre volume fraction. Hence, lf +lm
= vf . Similarly, this can be applied to the distance
travelled in the matrix. This facilitates Eq. 3.10 to be
Kh = vm Km + vf Kf (3.11)
In the case of this work, the fibre volume fraction is 10%. The absorption coefficient of
PLA and flax fibre is 80 1/m and 100 1/m respectively. Hence the calculated absorption
coefficient is 84 1/m.
The thermal history of the welding process at the interface is monitored using a thermo-
couple [119]. The temperature is measured with a custom-made programmable thermo-
couple with a Raspberry Pi miniature computer. The initial temperature is calibrated
with an infrared thermometer.
The temperature sensor is placed at the weld interface at the expected centre of the
weld line. A python-based algorithm is used to record the temperature during welding.
Two cases of welding are done, one with pure PLA plate as transparent part and the
second with PLA plate containing 10% short flax fibres. The plates have a thickness of
64
Heat transfer analysis
3 mm. The laser speed is set to 1 mm/s and the focus position is set to 70mm. The
welding is carried out in two cycles with a delay of 2 seconds between the cycles. Figure
3.13 shows the temperature profile at the centre point of the weld line for the two cases.
In the case of pure PLA, the maximum recorded temperature is approximately 200°C,
whereas for the semi-transparent composite it is 170°C. These tests are further performed
in the next chapter.
Temperature [℃]
Time Time
(a) (b)
Figure 3.13: Weld interface temperature for (a) Pure PLA and (b) PLA with 10% short
flax fibres.
Now, the numerical geometry is ready for ray tracing simulation. The ray tracing simula-
tions are carried with a focused beam having a focal angle as seen in the welding machine.
The beam is discretized into 200000 rays. Figure 3.14 shows an example of the ray direc-
tion at focus positions 72 mm and 70 mm when there is no obstruction in the path. All
the remaining parameters are the same as seen in the previous chapter.
To calculate the absorption coefficient in the semi-transparent part, modifications are
made to the ray tracing algorithm. Figure 3.15 is a schematic of the path of a ray that
is travelling through the thickness of the semi-transparent part. The distance travelled
by each ray in every macro-element is calculated during ray tracing simulation (d in
Equation 3.2). For each macro-element, the Beer-Lambert theory is applied to calculate
the attenuation of the laser intensity of every ray when it propagates through each macro-
65
3.4. Numerical method
Laser module
Focus position
72 mm 70 mm 65 mm
Laser beam
(Semi-) Transparent
part
Absorbent part
element.
Two profiles are created with the ray tracing simulation. A 2D plot of the weld
66
Heat transfer analysis
Laser rays
Transparent part 𝐼0
𝐼𝑎,2 𝑑1
𝐼1
𝐼2 𝑑2
Microstructures
𝐼𝑎,1
𝑑3
𝐼3 𝐼𝑎,3
Interface
Laser beam
Microstructures
Intensity W/m²
Transparent part
Interface
Figure 3.16: Contour plot of the laser intensity in all microstructures at the cross-section.
67
3.4. Numerical method
interface with a map of intensity and a 3D profile (Figure 3.16) of the energy absorbed
in the semi-transparent part. These two profiles are the input parameters for performing
heat transfer simulations.
COMSOL Multiphysics is a finite element solver for multiphysics simulations. This soft-
ware is used to model heat transfer during laser transmission welding in this work. The
heat transfer in the interface of the solid in COMSOL is used to model heat transfer in
solids by conduction, convection, and radiation. The interface solves the heat equation
with a time-dependent study [120]. In this work, laser transmission welding is modelled
by solving the heat equation and determining the temperature evolution in the weld zone
at the interface.
Semi-transparent
part
Absorbent part
A component geometry representing the semi-transparent part and the absorbent part
is constructed on COMSOL. The thickness of the two parts is 3 mm. The parts are 30 mm
long and 15 mm wide. A hexahedral mesh (Figure 3.17) is generated with additional fine-
ness in the weld zone. The material properties are user-defined, hence, the homogenised
values of Cp , ρ and kef f are input for semi-transparent material properties. The thermal
68
Heat transfer analysis
properties of the absorbent part are that of pure PLA, as seen in Table 3.1.
The initial value for temperature is taken as the default room temperature, 20°C.
Thermal insulation is applied on all the faces of the components except the interface. The
heat source input for the semi-transparent component is the intensity profiles generated
using ray tracing simulations. The intensity distribution data is applied as a function
depending on the volume of the semi-transparent part and the area of the weld interface.
This is done using Matlab live link linked to COMSOL. The simulation is carried out with
a moving heat source across the length of the geometry components at an input speed.
Thus, a time-dependent study is computed.
3.5 Results
Surface fusion marks as observed in the experimental study were simulated. Figure 3.18
shows the temperature distribution at the cross-section of the two components. Tempera-
ture above the melt temperature of PLA is seen at the surface of the semi-transparent part
in the first two cases. At a focus position of 72 mm, there seems to be a high probability
of material degradation. This was also observed in some experiments.
Though at lower lengths the burn mark can be completely omitted, it is to be noted
that the width of the weld line at the material interface is deeply affected by the focus
position. Figure 3.19 compares weld lines formed at high and low focus positions. With a
high focus position, a high portion of the laser energy is absorbed at the surface, resulting
in a surface burn mark and the meaning energy contributed to the interface weld line.
However, at low lengths, the laser energy has not been absorbed at the surface of the
semi-transparent part but is widely distributed at the weld interface. The line is wider
and there seem to be only a few regions where the polymer melted to cause interface
interdiffusion. The temperature distribution seems to be non-uniform in both cases.
3.6 Conclusion
Absorption of laser energy in a semi-transparent part was modelled in this chapter. During
welding of flax fibre reinforced PLA composites of 3mm thickness, a surface burn mark
69
3.6. Conclusion
Temperature
˚C
72 mm
70 mm
65 mm
70
Heat transfer analysis
was observed. The temperature field at melted regions at both interface and surface of
semi-transparent part was simulated numerically.
Among all the weld parameters, focus position seemed to have high impact on for-
mation of the burn mark. Hence this chapter focused on optimising the focus position
suitable for maximum energy transmission to the weld interface. A thermal model was
followed to simulate the absorption phenomenon by solving the 3D transient heat equa-
tion. As the semi-transparent part is heterogeneous, thermal and optical properties were
homogenised numerically. Temperature field in the assembly was determined with exper-
imental methods.
71
Chapter 4
Conductive heat transfer between the thermoplastic parts plays a vital role in defining
the assembly quality during laser transmission welding. During this step, within the
heat affected zone (HAZ), entanglement of polymer chains can be noticed leading to the
formation of interfacial bond aka weld. The weld strength can be expressed a function
of developed temperature field (T ), clamping pressure (p) and processing time (t) (see
Figure 4.1).
σ = f (T, p, t) (4.1)
The formation of a weld between the two parts involves two stages:
• The development of intimate contact between the semi-transparent part and the
absorbent part.
Interdiffusion can only occur after the intimate contact has been established. The opti-
mum quality of the weld line obtained at the interface of the two parts can be justified by
73
4.1. Degree of bonding
P, t Laser rays
Transparent part
T, t Interface
Absorbent part
P, t
T t P
Temperature Time Pressure
Figure 4.1: Inflience of physical parameters on weld line formation during laser transmis-
sion welding
developed mechanical strength, which is a function of the thermal history at the interface
of the two parts [121].
The objective of this chapter is to investigate the interfacial interactions between the
polymer chains at macromolecular level and linking them to the evolution of tempera-
ture in HAZ. Firstly, existing models based on isothermal and non-isothermal molecular
interdiffusion are presented and their advantages are discussed. A quasi non-isothermal
model is presented to predict the weld strength formation based on weld parameters. The
model is approached with temperature field information calculated through numerical
heat transfer simulations and measured through experimental procedures. The predicted
weld strength is validated with mechanical tests.
The quality of the bond formed by a weld line is determined by the degree of bonding
(Db ), which is calculated by considering two stages of the welding process and can be
expressed as:
74
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
where, Dic is the degree of intimate contact and Dau is the degree of autohesion (also
known as degree of healing Dh ). Bonding is assumed to be perfect when Db becomes
unity [21] and occurs only in the areas of weld interface that are in intimate contact [122].
The development of intimate contact is a function of the applied clamping pressure and
time, where as healing is a function of temperature history and time. Though the time
required to attain an intimate contact is much shorter than the time of autohesion, it
impacts greatly on the degree of bonding [123].
75
4.2. Intimate contact model
µmf
tic ∝ (4.3)
Papp
𝑝 𝑝
𝑡=0
𝑤0 𝑎0
𝑡>0
𝑤 𝑎
𝑏0
𝑏
Figure 4.2: Rectangular elements representing surface roughness before and after applying
clamping pressure.
The initial identical rectangles (at time t = 0) that represent the surface roughness
have a height of a0 , width b0 and are separated with a distance of w0 . The approach uses
volume conservation equation (equation 4.4) to express the evolution of the rectangles in
order to achieve an intimate contact upon applying clamping pressure [124].
V0 = a0 b0 = ab (4.4)
a0
a
Dic = w0 (4.6)
1+ b0
76
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
Further, the mass flow is considered laminar taking into account the viscosity of the
fibre-matrix mixture, followed by applying mass conservation law to control the volume
and as a function of the applied pressure [129].
In case of an isothermal and constant application of clamping pressure during the
process, Dic can be written as
1 n+1 n1
∂Dic n+1 n(n + 2) n A n Papp −n−3
=2 n Dic n (4.7)
∂t 2n + 1 (b0 + w0 )2 µmf
n
" n1 # 2n+3 1 n+1
2n+3 Papp 2n+1 (2n + 3)(n + 2) n A n
Dic (t) = Dic n
(0) + C t ,C = 2 n 2
µmf 2n + 1 (b0 + w0 )
" 2 # 15
1 5Papp w0 a0
Dic = w0 1 + 1+ t (4.8)
1 + b0 µmf b0 b0
A complete contact at the weld interface is achieved when the two materials are clamped
together with pressure Papp for a time t to achieve Dic = 1. The time required to reach a
full contact is,
2 " 5 #
µmf 1 b0 w0
tic = 1+ −1 (4.9)
5Papp 1 + wb00 a0 b0
Once an intimate contact is observed at the weld interface, the setup is eligible to underego
the bonding process namely, autohesion or healing. Figure 4.3 (c) shows that the region c
experiences healing with the migration of molecular chains at time tic . Hence, the degree
of healing can be expressed as
c
Dh = (4.10)
b
77
4.3. Interdiffusion at the weld interface
where c is the width of the area where molecular chain movement can be observed and b
is the width of the unit cell.
Interdiffusion process at the interface is influenced by the part geometry, material
thermal properties and welding process parameters [131]. When the laser beam energy
melts the two thermoplastic composites locally, allowing polymer chains to diffuse from
their original positions. Hence, the chains have an opportunity to interdiffuse within the
molten region for a time (tm ) However, the interdiffusion across the melt fronts is not
immediate. The polymer chains diffuse gradually beyond the interface, following a series
of disentanglements and entanglements. If the chain entanglement is the same as before
the melting, the original material strength is attained [132].
Initially, the diffusion mechanism between small molecules was well understood by the
Fickian law and the diffusion of the small molecules in the polymer network was explained
by a modified Fick’s law that included additional stresses due to the penetration of the
involved solvents. In cases of highly entangled chains, which is commonly seen in polymers
[133], the diffusion mechanism becomes more complicated.
The examination of highly entangled and thermally influenced interactions of polymer
chains is based on the chain dynamical theory known as reptation theory and is hypoth-
esized by deGennes [134]. An advanced reptation theory described by Wool et al. [21]
is generally accepted to describe the diffusion process of polymer chains. The premise of
reptation theory is that the polymer chain is confined to an imaginary tube (reptation
78
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
tube) which represents the surrounding entanglements of the other polymer chains. En-
tanglements resist polymer chains from each others movement. The model assumes that
the polymer chains can move freely within the tube but can leave the tube along the
curvilinear axis of the tube only. However, the reptation model does not consider realistic
welding processes where the temperature at the weld line drastically rises in time. Since
the model studies in detail the behaviour of the polymer chains, a steady, low healing tem-
perature at the interface has to be assumed [135]. Figure 4.4 summarises the procedure
of interdiffusion with respect to time.
Laser beam
t=0
Chain ends
t = t1
Temperature
Small chains
tm = Healing 𝒍𝟐
time (Quadratic distance
Reptation time
of interpenetration)
𝑅𝑔 2 (Tg or Tm)
t = tR 𝑡𝑅 = Rg
𝐷 (Radius of gyration)
Interface
Figure 4.4: Stages of interdiffusion of polymer chains during laser transmission welding
Figure 4.5 shows the timeline of polymer chain exiting the reptation tube where the
79
4.3. Interdiffusion at the weld interface
polymer chains require reptation time (tR ) to completely displace from the original tube.
Reptation time is expressed as
Rg2
tR = (4.11)
D
where, Rg is the radius of gyration and D is the interdiffusion coefficient.
According to the reptation theory, several characteristic time regimes can be delimited
considering the general motion behavior of the polymer chains. At short times known as
the Rouse relaxation time between entanglements (t < te ), a small motion takes place
on statistical segment without being subjected to topological constraints. Later, for te <
t < tr , which is the Rouse relaxation time for the whole chain, the chain is influenced by
the topological constraints of the tube and the motion is observed. For time tr < t < tR ,
which is the reptation relaxation time, the defects are equilibrated along the total length
of the tube and the polymer chain diffuses as whole along the fixed tube. Lastly, for
t > tR , the chain motion is governed by reptational dynamics that allow the chain to
renew the tube by escaping completely from the first tube without memory of its initial
conformation.
Reptation of the polymer chains also takes place across the weld lines and hence
contributes to the healing of weld lines [136]. Improper interdiffusion of molecular chains
across the interface leads to decrease of weld strength formation [137]. The effectiveness
of interdiffusion of polymer chains is determined be the average interpenetration distance
of the chains across the interface. At reptation time tR , the polymer chains interpenetrate
to mean square end-to-end distance of the polymer chain. The squared radius of gyration
is expressed in place of the mean square end-to-end distance of the polymer chain as it
facilitates to be measured experimentally. Hence, the chains should travel an average
distance equal to the radius of gyration [119]. The square radius of gyration (Rg2 ) is the
average squared distance of any point in the polymer chain from it’s center of mass, also
known as the moment of inertia.
Since the healing time can only be determined experimentally, the reptation time can
be used to obtain a prediction of the interfacial mechanical strength. The welding time
and the reptation time strongly depend on the temperature field in the weld zone: higher
temperatures give greater mobility of the polymer chain and therefore promote healing
80
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
< < ≤
𝑡=0 𝑡 = 𝑡1 𝑡 = 𝑡ℎ 𝑡 = 𝑡𝑅
at the interface. Under isothermal conditions, the bond strength is given by degree of
healing as the ratio of strength of the weld line (σw ) to the ultimate strength of the
material (σmax ) [132]:
σw
Dh = (4.12)
σmax
Studies [54, 137] have shown that the degree of healing is proportional to the fourth root
of time available for weld line to heal in the molten region. Hence,
σw 1
∝ t4 (4.13)
σmax
14
σw t
= (4.14)
σmax tR
At time tR , the polymer chains reorganise to erase the interface and weld strength is equal
to that of a pure material [135]. Reptation theory hence shows that the weld strength
development depends on keeping the weld interface equal to or above the melt temperature
(Tm ) for a time equal to or above tR .
According to Wool et. al [21], equation 4.12 is valid only for low molecular weights
M within the range defined using the critical entanglement molecular weight Mc : Mc <
M < 8Mc . At high molecular weights, the distance of interpenetration does not have
to reach radius of gyration Rg . Maximum bond strength can be achieved at a time th
(healing time) instead of tR . Healing time (th ) is see in Figure 4.5 (c) where, th ≤ tR .
Typically, engineering thermoplastics have higher molecular weights [22]. Hence, a
81
4.4. Interdiffusion under non-isothermal conditions
AR
tR = BR exp (4.16)
T
where, AR and BR are parameters to be determined experimentally.
Kim and Suh [137] defined an approach to calculate weld line strength based on molec-
ular interdiffusion. Here, the weld line factor is represented as a function of the the total
interface area in the weld line A0 and the area where interdiffusion has not yet taken place
A.
σw A
=1− (4.17)
σmax A0
The change in area (A) where interdiffusion has not taken place is determined by the
diffusion coefficient. According to Doi et al. [139], the coefficient of diffusion D is a
function of plateau modulus G0N [140], polymer density, ideal gas constant, temperature
T , mean square end-to-end distance Rg2 , mean molecular weight MW , critical entanglement
molecular weight Mc , and zero viscosity. This model determines the coefficient of diffusion
82
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
2 2
G0 Rg2
ρRT Mc (T )
D= N (4.18)
135 G0N MW 2
MW µ0,M (T )
Ezekoye et al. [135] noted that if the developed weld strength is plotted along side
interpenetration distance (l), the average diffusion distance in all the cases would collapse
to the same line and the ultimate strength would be achieved when l is equal to Rg . While
reptation theory considers that polymer chains to describe weld strength formation, it is
suitable to correlate welding parameters such as laser power and speed to bond strength
formation.
D̄ is used to calculate equivalent reptation time with equation 4.11. The degree of
healing at a time tm,max is written as,
14 14 √
tm,max tm,max D̄ l
Dh = = = (4.20)
t¯R
p
Rg2 1.1892 Rg
The degree of healing may also be defined by the length of the minor polymer chain (lm )
and the full length of the chain (L) (refer Figure 4.5).
1/2
σw lm
Dh = = (4.21)
σmax L
This equation is obtained by assuming that the factor of proportionality of the strength
with the square root of the minor chain length, lm , is the same for σ and σmax . Note
that Equation 4.21 applies to isothermal and nonisothermal conditions alike. However,
the growth of the minor chain length with time depends on the temperature history.
The temperature dependent reptation time tR can be calculated using the entangle-
83
4.4. Interdiffusion under non-isothermal conditions
Tn(t)
WL(x)
𝑫𝟏 𝑫𝟐 𝑫𝒏
ment molar mass MW , the density ρ, the zero viscosity µ0 , the ideal gas constant R, and
the temperature of the polymer Using the reptation model, temperature dependent tR
can be determined as
20Mc
tR (T ) = µ0 (4.22)
π 2 RT ρ
1. Reptation theory describes the formation of weld strength by examining the quadratic
distance of inter-diffusion (⟨l⟩) of polymer chains across the weld interface.
2. During reptation, ⟨l⟩2 = Rg [141] and weld strength (σ) = strength of pure material
(σm ).
4. Maintaining melt temperature at interface for a time (t) longer than tR is unneces-
sary as this will not result in additional strength formation.
It should be noted that the models mentioned earlier in this chapter were developed for
amorphous polymers. It has been identified by Akkerman et al. [142] that, in literature,
there appears to be no consensus on how to model the healing process of semi-crystalline
84
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
polymers. The non-isothermal healing model developed by Yang and Pitchumani [127]
only applies for amorphous polymers and is, therefore, only applicable for semi-crystalline
polymers in a melted state and can continue until the polymer starts to crystallize. Nev-
ertheless, in some studies the healing model for amorphous polymers was directly applied
for semi-crystalline polymers without considering the effect of the crystalline structure
above the glass transition temperature [130, 143]. It is safe to assume that healing of
semi-crystalline polymers can occur only above the melting temperature [122, 22].
85
4.5. Proposed model to predict the weld strength
1/4
in this case, tm . At time t = 0, Dh = 0 which is at the start of the process. A straight
solid line passing through the origin and the formerly plotted point can be drawn because
it denotes a best linear regression fit of the data [22]. The healing time is obtained from
−1/4
the slope of the linear variations, which equals th , from the intercept of the solid line
1/4
with Dh = 1, which equals th . Yang et. al validated this approach by fitting the data
to an Arrhenius relationship with respect to temperature. The healing time determined
in each case can be plotted against temperature. An interpolation of this data should be
sufficient to determine the required healing time as a function of temperature.
The optimisation of the welding process in the present work concerns short flax fibre
reinforced thermoplastic composites. Due to impacting factors such as absorption and
diffusion in the semi-transparent part, the current model cannot directly relate the degree
of healing to welding speed. Therefore, a knowledge of temperature (Ts ) at the weld
interface influenced by the semi-transparent part, is needed.
In the previous chapters of this thesis, effects of fibres in the semi-transparent part were
addressed and temperature at the weld interface was predicted. This data also exhibits
tm . To remind, the temperature measurements were verified experimentally. From the
previously gathered data of pure PLA, the value of Ts can be interpolated to find the
relevant healing time (th,s ) required.
This model is verified in the following sections. Firstly, temperature and heal time
information of pure PLA with respect to welding speed is gathered experimentally. Then,
temperature development at HAZ is determined with two approaches: experimental and
numerical. The weld line is discredited sparsely. In the experimental approach, tempera-
ture is measured at each section of the weld with a thermocouple. On the other hand, the
LTW process is numerically simulated and temperature at the same locations are noted.
The explained model is then approached. These results will be compared with mechanical
tests.
86
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
The application of the proposed model firstly requires a summary of temperature data as
a function of welding speed. For this, it is required to determine suitable weld parameters
that form a bond. This work continues to use pure PLA and short flax fibre reinforced
PLA composite as transparent and semi-transparent parts, and PLA with 1% MB black
colorant as the absorbent part. Laser transmission welding simulations of non-reinforced
PLA plates is numerically performed on COMSOL Multiphysics. The laser beam is
circular with Gaussian distribution of intensity. Weld parameters such as speed and
power are varied while keeping the focus position constant. Thermal properties of PLA
are allocated to both the components, where as coefficient of absorption is input only to
the absorbent part. Adiabatic boundary conditions are applied. Maximum temperature
reached at the center of the HAZ is recorded. This allows to relate the weld parameters to
developed temperature with a contour plot as shown in Figure 4.7. Now, the plot can be
narrowed down further with the melt temperature range of 155-170°C for PLA (IngeoTM
Biopolymer 4032D) [43].
10
1 5515
5 155℃ Melt temperature
9
155 170 170℃ (𝑇𝑚 ) range of PLA
55
1551
𝑇 < 𝑇𝑚
8 1 70
170 7 0
1
155
Laser velocity [mm/s]
5
7 15 155
170171070
15 5
6
170
170 𝑇 > 𝑇𝑚
5
30 0300
300 300℃
4
300 300
300
300
3 00
30 0
3
300
Figure 4.7: Contour plot of temperature with respect to laser speed and power
87
4.5. Proposed model to predict the weld strength
Transparent part
Absorbent part
Weld zone
Weld line
15 mm
30 mm
88
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
1/4
𝑡ℎ
Figure 4.9: Degree of healing vs. melt time for weld speed of 7mm/s
10 mm/s
12 mm/s
11 mm/s
8 mm/s
9 mm/s
7 mm/s
𝐷ℎ
𝑡 0.25
Figure 4.10: Degree of healing with respect to melt time for all cases of laser movement
speed.
89
4.5. Proposed model to predict the weld strength
Table 4.1: Welding process parameters for lap shear specimens and HAZ temperature
measurements
1/4
Time 𝑡𝑅
Figure 4.11: Exponential plot of the calculated reptation time as a function of maximum
temperature.
To verify the model with a semi-transparent part, flax fibre reinforced PLA composite,
temperature data at the weld line is collected. Due to the heterogeneity of the fibre distri-
bution in the semi-transparent part, a non-uniform temperature distribution is expected
along the weld line. To record these temperatures, the length of the weld line is discretized
90
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
Table 4.2: Experimental data to calculate reptation time of PLA at various temperatures
at interface
into five sections (see Figure 4.12) and temperature is measured with five thermocouples.
The laser power is set to 46W. The focus position is set at 68mm while welding speed to
2mm/s. From the previously conducted experiments, it is evident that the low welding
speed will not degrade the polymer at the weld interface due to diffusion and absorption
properties in the semi-transparent part. A welding speed as low as 2mm/s may, in fact,
cause a non-uniform weld line. The recorded temperature curves are seen in Figure 4.13.
1 2 3 4 5
Weld line
Further calculations are made using MATLAB 2021b. The temperature at each loca-
tion is interpolated in the previously processed data (Figure 4.11) and the corresponding
healing time is determined. For location 1, the temperature T1 = 184°C correlates with
1/4 1/4
healing time th,1 = 1.83s.th,1 can be plotted against degree of healing at Dh,1 = 1. The
91
4.5. Proposed model to predict the weld strength
1 4
2
3 155° C
5
Figure 4.13: Temperature profile measured at five locations of weld line at interface of
flax fibre reinforced PLA composite and pure PLA plates.
92
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
melt time at this location (tm,1 ) is recorded as 0.75 seconds. Hence the degree of healing
is calculated to be 0.5079.
1/4
𝑡ℎ,1
𝐷ℎ
1/4
𝑡𝑚,1
1
Time, 𝑡 4
The degree of healing is calculated for the remaining sections of the weld line. The
summation of all sections is the eventual average strength of the weld line.
Pn
x=1 σx
σweldline = (4.24)
n
where n is the number of sections in the weld line, n = 1 to 5
93
4.7. Results and discussion
22
11 33
4 5
5
155° C
Temperature ℃
𝑡𝑚,2
Time
94
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
Tensile tests were performed on Instron 1185 Universal Testing Machine at cross-head
speed of 1mm/min. The tensile strength is obtained using Equation 1.8. Area of HAZ
is measured after the mechanical tests. An example is seen in Figure 4.17. The average
weld strength determined in all approaches is shown in Table 4.3.
2 mm
15.5 mm
Figure 4.17: Dimensions of weld line at weld interface. Weld line produced at laser speed
of 2mm/s.
The curves are fitted with a smoothing spline and a smoothing parameter of 0.96.
The following observations were made:
95
4.7. Results and discussion
96
Study of molecular interdiffusion in a polymer-polymer weld line: Optimisation of laser
welding process
1. Degree of intimate contact Dic is considered to have approached unity. In the laser
welding machine at IMT Nord europe, the clamping pressure is controlled auto-
matically. Because the plates were manufactured by injection molding, the surface
roughness may be comparatively low, but, it cannot be completely eliminated. The
weld strength calculated with simulated inputs could be impacted as complete con-
tact is observed at the weld interface. On the other hand, the experimental input
results may be impacted, especially while measuring HAZ temperature. The ther-
mocouple hinders the occurrence of an intimate contact.
2. The current calculations are made with one cycle of laser welding. A higher weld
strength can be achieved with multiple cycles as it increases melt time.
3. Power or focus position can replace welding speed in the model. This allows a more
complex and accurate prediction of weld strength as reference data increases for
interpolation.
4. The approach to the model is both isothermal and non-isothermal at different steps.
In the above calculations, the temperature T is the maximum recorded temperature,
but the melt time tm is not obtained at T . A more accurate result can be obtained
by discretizing the temperature peak above Tm and calculating Dh at every element
with isothermal approach. A summation of weld strength of all elements will be the
total weld strength at a perticular location.
5. The proposed model is validated with experimental and simulation inputs. Another
disadvantage of experimental inputs is the temperature measurements cannot be
made at many locations on the weld line. Temperature measurements from heat
transfer simulations can provide better information on temperature development,
especially considering the absorption and scatter phenomena.
4.8 Conclusion
The formation of a weld was discussed following the impact of physical parameters on the
weld strength. Welding parameters and molecular interdiffusion are linked to temperature
97
4.8. Conclusion
developed at the heat affected zone. The degree of bonding was calculated by considering
intimate contact model and the autohesion.
Models to calculate the degree of healing at the weld interface were analysed at isother-
mal and non-isothermal conditions. The time required to form a good weld at various
laser welding speed was calculated by keeping the laser power constant. A model was
proposed based on existing healing models to predict the weld strength along the weld
line length.
Weld line strength was predicted by verifying the proposed model with interface tem-
perature data produced experimentally and numerically. The degree of healing at multiple
sections of the weld line was calculated in each case and averaged. Lap-joint specimens
were manufactured with semi-transparent plate of flax fibre reinforced PLA composite
and absorbent part made of PLA and 1% MB black colorant. Tensile test results of the
lap-joint specimens were used to validate the predicted strength of the weld line.
98
Chapter 5
Conclusions
The process of laser transmission welding of natural fibre reinforced thermoplastic com-
posites was presented in this dissertation. Firstly, the importance of implementation of
assembly in bio-composites was studied. The challenges of short fibre inclusions in the
semi-transparent part was investigated. The possibilities of numerically modelling the
welding process were also studied.
Following this, the influence of fibre orientation on laser beam diffusion was determined
numerically and experimentally. A method to generate 3D numerical geometries was
developed by considering physical fibre morphology in an injection moulded composite
material. A ray tracing simulation was implemented to analyse the diffusion of laser
rays in a heterogeneous medium with respect to fibre orientation. It was concluded that
fibre orientation has a large influence on lase beam diffusion and it will impact the laser
intensity distribution at the weld interface.
The study of impact of fibre orientation on light diffusion allowed to generate com-
plex numerical geometries with multi-layer orientation distribution extracted from fibre
orientation tensors.
Absorption in the semi-transparent part was studied by modelling the fusion mark
caused due to surface energy absorption. A method to numerically calculate absorption
coefficient of the semi-transparent part was presented. The influence of fibre morphology
99
on absorption was studied. It was concluded that the fusion mark was caused due to one
of the main welding parameters, focus position. The temperature distribution at the weld
interface as a function of weld parameters was studied. Optimal welding parameters to
such as welding speed and power required to heat the weld interface to a melt temperature
were determined with heat transfer simulations.
The thermal properties such as density, heat capacity and thermal conductivity were
homogenised numerically for the short flax fibre reinforced PLA composites which is
primarily used in the investigations of this dissertation.
Also, an experimental procedure to measure interface weld temperature using ther-
mocouple was demonstrated. The measurements were made for weld zones for pure PLA
plates and PLA flax composites. The affect of welding process parameters, especially
laser movement speed was studied.
A model has been proposed to determine the degree of healing based on existing
isothermal and non-isothermal adhesion approaches. The process of LTW was optimised
by linking weld parameters to molecular interdiffusion at the weld zone. Reptation time
was calculated using mechanical tests and as a function of laser welding speed. A pre-
diction of weld strength was made based on the interface temperature that was measured
experimentally and numerically. The results were successfully validated with mechanical
tests.
Perspectives
Generating 3D numerical structures was one of the most challenging tasks of this project.
Moreover, the accuracy of the numerical simulations can only be reached with construction
of the semi-transparent part without periodicity. Optimising the algorithm to generate 3D
microstructures with computational cost effective methods would be interesting. Fibre
orientation distribution of the numerical geometries were generated from fourth order
orientation tensors. More accuracy in the orientation can be found with sixth-order
tensors.
Apart from this, as the algorithm is capable of generating polyhedrons to represent
fibre inclusions, it would be interesting to modify the inclusion shapes to represent com-
100
Perspectives
plex bundles of short natural fibres. The influence of fibre shape on laser diffusion and
temperature distribution can be further studied.
The calculation of absorption coefficient in the microstructure is done at a macro scale.
For more accuracy, it can be calculated at a micro scale for each micro element in the
semi-transparent part. This could be computationally expensive but the periodicity of
the the microstructures might be an advantage.
Lap shear tests to calculate weld strength has a disadvantage of peel effect on the weld
line. To determine purely the shear strength of the weld line, other mechanical testing
methods must be attempted. It is important to notice that the length and number of
weld line plays a crucial role in strength formation.
In chapter 4, the degree of healing is calculated for specimens with a change in welding
speed. The temperature in this case is a function of weld speed. The model could be
further developed to calculate degree of healing as a function of laser power.
To make the welding of composites accessible to the manufacturing industry, applica-
tions can be created with the proposed model of weld strength calculation while taking
into account the welding parameters and material proprieties. It is possible to include
the manufacturing process to predict the fibre orientation in all regions of the part and
analyse the laser diffusion in the weld area.
101
Bibliography
[3] AK Bledzki and Jochen Gassan. “Composites reinforced with cellulose based fi-
bres”. In: Progress in polymer science 24.2 (1999), pp. 221–274.
[5] Rainier Zah et al. “Curauá fibers in the automobile industry–a sustainability as-
sessment”. In: Journal of cleaner production 15.11-12 (2007), pp. 1032–1040.
[7] Vijay K Stokes. “Experiments on the hot-tool welding of three dissimilar thermo-
plastics”. In: Polymer 39.12 (1998), pp. 2469–2477.
[8] N Mendes et al. “Morphology and strength of acrylonitrile butadiene styrene welds
performed by robotic friction stir welding”. In: Materials & Design 64 (2014),
pp. 81–90.
103
Bibliography
[10] Negin Amanat, Natalie L James, and David R McKenzie. “Welding methods for
joining thermoplastic polymers for the hermetic enclosure of medical devices”. In:
Medical engineering & physics 32.7 (2010), pp. 690–699.
[11] Rolf Klein. Laser welding of plastics: materials, processes and industrial applica-
tions. John Wiley & Sons, 2012.
[12] Ian Jones. “Laser welding of plastics”. In: Handbook of Laser Welding Technologies.
Elsevier, 2013, 280–301e.
[13] William M Steen and Jyotirmoy Mazumder. Laser material processing. springer
science & business media, 2010.
[14] Michael J Troughton. Handbook of plastics joining: a practical guide. William An-
drew, 2008.
[16] Val Kagan, Robert Bray, and Al Chambers. “Forward to better understanding
of optical characterization and development of colored polyamides for the infra-
red/laser welding: part I-efficiency of polyamides for infra-red welding”. In: Journal
of reinforced plastics and composites 22.6 (2003), pp. 533–547.
[17] André Chateau Akué Asséko et al. “Laser transmission welding of composites-Part
A: Thermo-physical and optical characterization of materials”. In: Infrared Physics
& Technology 72 (2015), pp. 293–299.
[18] André Chateau Akué Asséko et al. “Analytical and numerical modeling of light
scattering in composite transmission laser welding process”. In: International Jour-
nal of Material Forming 8.1 (2015), pp. 127–135.
104
Bibliography
[20] James D Van de Ven and Arthur G Erdman. “Laser transmission welding of ther-
moplastics—Part I: Temperature and pressure modeling”. In: (2007).
[21] Richard P Wool, B-L Yuan, and OJ McGarel. “Welding of polymer interfaces”. In:
Polymer Engineering & Science 29.19 (1989), pp. 1340–1367.
[23] Luis FFF Gonçalves et al. “Laser welding of thermoplastics: An overview on lasers,
materials, processes and quality”. In: Infrared Physics & Technology 119 (2021),
p. 103931.
[24] HJ Silvers Jr and S Wachtell. “Perforating, welding and cutting plastic films with
a continuous CO2 laser’. PA State University, Eng”. In: Proc. 1970, pp. 88–97.
[25] Elhem Ghorbel, Giuseppe Casalino, and Stéphane Abed. “Laser diode transmis-
sion welding of polypropylene: Geometrical and microstructure characterisation of
weld”. In: Materials & Design 30.7 (2009), pp. 2745–2751.
[26] Foram Dave et al. “Laser transmission welding of semi-crystalline polymers and
their composites: A critical review”. In: Polymers 13.5 (2021), p. 675.
[28] B Acherjee et al. “Selection of process parameters for optimizing the weld strength
in laser transmission welding of acrylics”. In: Proceedings of the Institution of Me-
chanical Engineers, Part B: Journal of Engineering Manufacture 224.10 (2010),
pp. 1529–1536.
[29] Hyun Myung Shin and Hae Woon Choi. “Design of energy optimization for laser
polymer joining process”. In: The International Journal of Advanced Manufactur-
ing Technology 75.9 (2014), pp. 1569–1576.
[30] Julian Brodhun, David Blass, and Klaus Dilger. “Laser transmission welding of
thermoplastic fasteners: Influence of temperature distribution in a scanning based
process”. In: Procedia CIRP 74 (2018), pp. 533–537.
105
Bibliography
[33] Malladi Nagalakshmaiah et al. “Biocomposites: Present trends and challenges for
the future”. In: Green composites for automotive applications (2019), pp. 197–215.
[34] Eshetie Kassegn, Frederik Desplentere, and Temesgen Berhanu. “Mechanical Prop-
erties of Short Sisal Fiber Reinforced Poly Lactic Acid (PLA) Processed by Injec-
tion Molding”. In: Branna Journal of Engineering and Technology (2019), pp. 20–
36.
[35] Pierre Ouagne et al. “Fibre extraction from oleaginous flax for technical tex-
tile applications: influence of pre-processing parameters on fibre extraction yield,
size distribution and mechanical properties”. In: Procedia Engineering 200 (2017),
pp. 213–220.
[36] Marko Hyvärinen, Rowshni Jabeen, and Timo Kärki. “The modelling of extrusion
processes for polymers—A review”. In: Polymers 12.6 (2020), p. 1306.
[37] Narendra Reddy and Yiqi Yang. “Non-traditional lightweight polypropylene com-
posites reinforced with milkweed floss”. In: Polymer International 59.7 (2010),
pp. 884–890.
[39] Maya Jacob John and Sabu Thomas. “Biofibres and biocomposites”. In: Carbohy-
drate polymers 71.3 (2008), pp. 343–364.
[40] Andre Chateau Akue Asseko et al. “Effect of the developed temperature field on
the molecular interdiffusion at the interface in infrared welding of polycarbonate
composites”. In: Composites Part B: Engineering 97 (2016), pp. 53–61.
106
Bibliography
[41] Myriam Dauphin and Benoit Cosson. “Modeling of thermoplastic composites laser
welding-a ray tracing method associated to thermal simulation”. In: AIP Confer-
ence Proceedings. Vol. 1769. 1. AIP Publishing LLC. 2016, p. 020010.
[42] Mariana Ilie et al. “Laser beam scattering effects in non-absorbent inhomogenous
polymers”. In: Optics and Lasers in Engineering 45.3 (2007), pp. 405–412.
[44] Mingliang Chen, Gene Zak, and Philip J Bates. “Effect of carbon black on light
transmission in laser welding of thermoplastics”. In: Journal of Materials Process-
ing Technology 211.1 (2011), pp. 43–47.
[45] V Wippo et al. “Advanced laser transmission welding strategies for fibre reinforced
thermoplastics”. In: Physics Procedia 56 (2014), pp. 1191–1197.
[47] John R Howell et al. Thermal radiation heat transfer. CRC press, 2020.
[48] Christine A Butler et al. “An analysis of mechanisms governing fusion bonding
of thermoplastic composites”. In: Journal of Thermoplastic Composite Materials
11.4 (1998), pp. 338–363.
[49] Georges Ayoub et al. “Modelling large deformation behaviour under loading–
unloading of semicrystalline polymers: application to a high density polyethylene”.
In: International Journal of Plasticity 26.3 (2010), pp. 329–347.
[50] Andrzej Pawlak, Andrzej Galeski, and Artur Rozanski. “Cavitation during defor-
mation of semicrystalline polymers”. In: Progress in polymer science 39.5 (2014),
pp. 921–958.
107
Bibliography
[52] K Schmidt-Rohr and HW Spiess. “Chain diffusion between crystalline and amor-
phous regions in polyethylene detected by 2D exchange carbon-13 NMR”. In:
Macromolecules 24.19 (1991), pp. 5288–5293.
[53] Anurag Pandey, Akihiko Toda, and Sanjay Rastogi. “Influence of amorphous com-
ponent on melting of semicrystalline polymers”. In: Macromolecules 44.20 (2011),
pp. 8042–8055.
[54] Young Hwa Kim and Richard P Wool. “A theory of healing at a polymer-polymer
interface”. In: Macromolecules 16.7 (1983), pp. 1115–1120.
[55] RPa Wool and KM O’connor. “A theory crack healing in polymers”. In: Journal
of applied physics 52.10 (1981), pp. 5953–5963.
[56] RP Wool and KM O’Connor. “Time dependence of crack healing”. In: Journal of
polymer science: Polymer letters edition 20.1 (1982), pp. 7–16.
[57] PJ Bates, TB Okoro, and M Chen. “Thermal degradation of PC and PA6 during
laser transmission welding”. In: Welding in the World 59.3 (2015), pp. 381–390.
[59] LS Mayboudi et al. “A 3-D thermal model for spot laser transmission welding of
thermoplastics”. In: Welding in the World 51.1 (2007), pp. 74–78.
[60] Manfred Geiger, T Frick, and Michael Schmidt. “Optical properties of plastics
and their role for the modelling of the laser transmission welding process”. In:
Production Engineering 3.1 (2009), pp. 49–55.
[62] Mirko Aden. “Influence of the laser-beam distribution on the seam dimensions for
laser-transmission welding: a simulative approach”. In: Lasers in Manufacturing
and Materials Processing 3.2 (2016), pp. 100–110.
108
Bibliography
[63] Andre Chateau Akue Asseko et al. “Thermal modeling in composite transmission
laser welding process: light scattering and absorption phenomena coupling”. In:
ESAFORM 2014-17th Conference of the European Scientific Association on Ma-
terial forming. Vol. 611. 2014, p–1560.
[64] Benoit Cosson et al. “3D modeling of thermoplastic composites laser welding
process–A ray tracing method coupled with finite element method”. In: Optics
& Laser Technology 119 (2019), p. 105585.
[65] Bappa Acherjee et al. “Optimal process design for laser transmission welding of
acrylics using desirability function analysis and overlay contour plots”. In: Inter-
national Journal of Manufacturing Research 6.1 (2011), pp. 49–61.
[66] LS Mayboudi et al. “Infrared observations and finite element modeling of a laser
transmission welding process”. In: Journal of Laser Applications 21.3 (2009), pp. 111–
118.
[67] Bappa Acherjee et al. “Application of artificial neural network for predicting weld
quality in laser transmission welding of thermoplastics”. In: Applied soft computing
11.2 (2011), pp. 2548–2555.
[69] L Yan et al. “Characterizing Temporary Bonding Adhesives Using a Wedge Test”.
In: 12th International Conference on Fracture 2009, ICF-12. 2009, pp. 6347–6355.
[70] J Renart et al. “Mode I fatigue behaviour and fracture of adhesively-bonded fibre-
reinforced polymer (FRP) composite joints for structural repairs”. In: Fatigue and
fracture of adhesively-bonded composite joints. Elsevier, 2015, pp. 121–147.
[71] Jakub Rzeczkowski. “An experimental analysis of the end-notched flexure com-
posite laminates beams with elastic couplings”. In: Continuum Mechanics and
Thermodynamics 33.6 (2021), pp. 2331–2343.
109
Bibliography
[73] France Chabert et al. “Transmission Laser Welding of Polyamides: Effect of Process
Parameter and Material Properties on the Weld Strength”. In: Procedia Manufac-
turing 47 (2020), pp. 962–968.
[74] Mariana Ilie et al. “Diode laser welding of ABS: Experiments and process model-
ing”. In: Optics & Laser Technology 41.5 (2009), pp. 608–614.
[75] André Chateau Akué Asséko et al. “Laser transmission welding of composites–Part
B: Experimental validation of numerical model”. In: Infrared Physics & Technology
73 (2015), pp. 304–311.
[76] Fransisco Folgar and III Charles L. Tucker. “Orientation Behavior of Fibers in
Concentrated Suspensions”. In: Journal of Reinforced Plastics and Composites 3.2
(1984), pp. 98–119.
[78] Benoit Cosson, Mylène Deléglise, and Wolfgang Knapp. “Numerical analysis of
thermoplastic composites laser welding using ray tracing method”. In: Composites
Part B: Engineering 68 (2015), pp. 85–91.
[80] MJ Folkes and DAM Russell. “Orientation effects during the flow of short-fibre
reinforced thermoplastics”. In: Polymer 21.11 (1980), pp. 1252–1258.
[81] Michel Vincent et al. “Description and modeling of fiber orientation in injection
molding of fiber reinforced thermoplastics”. In: Polymer 46.17 (2005), pp. 6719–
6725.
[82] Brent E VerWeyst et al. “Fiber orientation in 3-D injection molded features”. In:
International Polymer Processing 14.4 (1999), pp. 409–420.
110
Bibliography
[83] Radmir Karamov et al. “Micro-CT based structure tensor analysis of fibre orienta-
tion in random fibre composites versus high-fidelity fibre identification methods”.
In: Composite Structures 235 (2020), p. 111818.
[87] Suresh G Advani and Charles L Tucker III. “The use of tensors to describe and
predict fiber orientation in short fiber composites”. In: Journal of rheology 31.8
(1987), pp. 751–784.
[88] André Chateau Akué Asséko et al. “Numerical analysis of effective thermal conduc-
tivity of plastic foams”. In: Journal of Materials Science 51.20 (2016), pp. 9217–
9228.
[90] Karen Soete et al. “Variability of flax fibre morphology and mechanical properties
in injection moulded short straw flax fibre-reinforced PP composites”. In: Journal
of Composite Materials 51.23 (2017), pp. 3337–3349.
[91] Joaquim S Cintra Jr and Charles L Tucker III. “Orthotropic closure approxi-
mations for flow-induced fiber orientation”. In: Journal of Rheology 39.6 (1995),
pp. 1095–1122.
111
Bibliography
[92] Du Hwan Chung and Tai Hun Kwon. “Improved model of orthotropic closure
approximation for flow induced fiber orientation”. In: Polymer composites 22.5
(2001), pp. 636–649.
[93] DA Jack and DE Smith. “An invariant based fitted closure of the sixth-order
orientation tensor for modeling short-fiber suspensions”. In: Journal of Rheology
49.5 (2005), pp. 1091–1115.
[94] Suresh G Advani and Charles L Tucker III. “Closure approximations for three-
dimensional structure tensors”. In: Journal of Rheology 34.3 (1990), pp. 367–386.
[95] Delphine Dray, Pierre Gilormini, and Gilles Régnier. “Comparison of several clo-
sure approximations for evaluating the thermoelastic properties of an injection
molded short-fiber composite”. In: Composites Science and Technology 67.7-8 (2007),
pp. 1601–1610.
[98] Benoit Cosson et al. “Infrared heating stage simulation of semi-transparent media
(PET) using ray tracing method”. In: International Journal of Material Forming
4.1 (Mar. 2011), pp. 1–10.
[99] Mirko Aden, Andreas Roesner, and Alexander Olowinsky. “Optical characteriza-
tion of polycarbonate: Influence of additives on optical properties”. In: Journal of
Polymer Science Part B: Polymer Physics 48.4 (2010), pp. 451–455.
[100] Juan Manuel Vazquez-Martinez et al. “Evaluation of the joining response of biodegrad-
able polylactic acid (PLA) from fused deposition modeling by infrared laser irra-
diation”. In: Polymers 12.11 (2020), p. 2479.
112
Bibliography
[101] David Grewell, Paul Rooney, and Val A Kagan. “Relationship between optical
properties and optimized processing parameters for through-transmission laser
welding of thermoplastics”. In: Journal of Reinforced plastics and Composites 23.3
(2004), pp. 239–247.
[104] Wolfgang Knapp et al. “Laser-bonding of long fiber thermoplastic composites for
structural assemblies”. In: Physics Procedia 5 (2010), pp. 163–171.
[105] Janis Sliseris, Libo Yan, and Bohumil Kasal. “Numerical modelling of flax short
fibre reinforced and flax fibre fabric reinforced polymer composites”. In: Composites
part B: engineering 89 (2016), pp. 143–154.
[106] Benoit Cosson, André Chateau Akué Asséko, and Myriam Dauphin. “A non-
destructive optical experimental method to predict extinction coefficient of glass
fibre-reinforced thermoplastic composites”. In: Optics & Laser Technology 106
(2018), pp. 215–221.
[107] I. Jones. “13 - Transmission laser welding strategies for medical plastics”. In: Join-
ing and Assembly of Medical Materials and Devices. Ed. by Y. Zhou and Mark D.
Breyen. Woodhead Publishing Series in Biomaterials. Woodhead Publishing, 2013,
344–371e. isbn: 978-1-84569-577-4.
[108] Marcus Warwick and Marcus Gordon. “Application studies using through-transmission
laser welding of polymers”. In: Joining Plastics (2006), pp. 25–26.
[110] A Roesner, M Aden, and A Olowinsky. “Laser Transmission Welding Under Spe-
cial Consideration of Scattering”. In: Proceedings of the Fifth International WLT-
Conference on Lasers in Manufacturing, Munich. 2009, p. 2.
113
Bibliography
[111] Suthaphat Kamthai and Rathanawan Magaraphan. “Thermal and mechanical prop-
erties of polylactic acid (PLA) and bagasse carboxymethyl cellulose (CMCB) com-
posite by adding isosorbide diesters”. In: AIP Conference Proceedings 1664.1 (2015),
p. 060006.
[112] Oldrich Zmeskal et al. “Thermal properties of samples prepared from polylactic
acid by 3D printing”. In: AIP Conference Proceedings 2305.1 (2020), p. 020022.
[113] Romain Léger et al. “Modeling natural fiber composites microstructures for pre-
dicting their water aging”. In: 18th European Conference on Composites Materials.
2018.
[115] Zhidong Han and Alberto Fina. “Thermal conductivity of carbon nanotubes and
their polymer nanocomposites: A review”. In: Progress in polymer science 36.7
(2011), pp. 914–944.
[118] Hanna-Riitta Kymäläinen et al. “Quality of Linum usitatissimum L.(flax and lin-
seed) and Cannabis sativa L.(fibre hemp) during the production chain of fibre raw
material for thermal insulations”. In: (2004).
[119] LJ Bastien and JW Gillespie Jr. “A non-isothermal healing model for strength
and toughness of fusion bonded joints of amorphous thermoplastics”. In: Polymer
Engineering & Science 31.24 (1991), pp. 1720–1730.
114
Bibliography
[122] Woo Il Lee and George S Springer. “A model of the manufacturing process of
thermoplastic matrix composites”. In: Journal of composite materials 21.11 (1987),
pp. 1017–1055.
[125] Khaled Yassin and Mehdi Hojjati. “Processing of thermoplastic matrix compos-
ites through automated fiber placement and tape laying methods: A review”. In:
Journal of Thermoplastic Composite Materials 31.12 (2018), pp. 1676–1725.
[126] AC Loos and PH Dara. “Modeling the curing process of thick-section autoclave
cured composites”. In: Proceedings of the 2nd Annual Review of the Center for
Composite Materials and Structures (1985).
[127] F Yang and R Pitchumani. “A fractal Cantor set based description of interlami-
nar contact evolution during thermoplastic composites processing”. In: Journal of
materials science 36.19 (2001), pp. 4661–4671.
[128] Gilles Regnier and Steven Le Corre. “Modeling of thermoplastic welding”. In:
(2016).
[131] D Grewell and A Benatar. “Coupled temperature, diffusion and squeeze flow model
for interfacial healing predictions”. In: Annual Technical Conference for the Soci-
ety of Plastic Engineers Proceedings, Society of Plastic Engineers, Brookfield, CT.
2006.
115
Bibliography
[132] Jakob Onken, Steffen Verwaayen, and Christian Hopmann. “Evaluation of heal-
ing models to predict the weld line strength of the amorphous thermoplastic
polystyrene by injection molding simulation”. In: Polymer Engineering & Science
61.3 (2021), pp. 754–766.
[133] Huagui Zhang, Khalid Lamnawar, and Abderrahim Maazouz. “Rheological mod-
eling of the diffusion process and the interphase of symmetrical bilayers based
on PVDF and PMMA with varying molecular weights”. In: Rheologica acta 51.8
(2012), pp. 691–711.
[135] OA Ezekoye et al. “Polymer weld strength predictions using a thermal and polymer
chain diffusion analysis”. In: Polymer Engineering & Science 38.6 (1998), pp. 976–
991.
[136] H H Kausch and M Tirrell. “Polymer interdiffusion”. In: Annual Review of Ma-
terials Research 19 (1989), pp. 341–377.
[137] Sang-Gook Kim and Nam P Suh. “Performance prediction of weldline structure in
amorphous polymers”. In: Polymer Engineering & Science 26.17 (1986), pp. 1200–
1207.
[138] ULF Gedde. Polymer physics. Springer Science & Business Media, 1995.
[139] Masao Doi, Samuel Frederick Edwards, and Samuel Frederick Edwards. The theory
of polymer dynamics. Vol. 73. oxford university press, 1988.
[140] Chenyang Liu et al. “Evaluation of different methods for the determination of
the plateau modulus and the entanglement molecular weight”. In: Polymer 47.13
(2006), pp. 4461–4479.
[141] Joseph Bartolai, Timothy W Simpson, and Renxuan Xie. “Predicting strength of
thermoplastic polymer parts produced using additive manufacturing”. In: 2016 In-
ternational Solid Freeform Fabrication Symposium. University of Texas at Austin.
2016.
116
Bibliography
[142] Remko Akkerman, Mark Bouwman, and Sebastiaan Wijskamp. “Analysis of the
thermoplastic composite overmolding process: Interface strength”. In: Frontiers in
Materials 7 (2020), p. 27.
[143] Arthur Levy, Steven Le Corre, and Arnaud Poitou. “Ultrasonic welding of thermo-
plastic composites: a numerical analysis at the mesoscopic scale relating processing
parameters, flow of polymer and quality of adhesion”. In: International Journal of
Material Forming 7.1 (2014), pp. 39–51.
117
Abstract
Laser transmission welding of thermoplastics requires the optimisation of interfacial adhesion at the
weld joint. In this regard, the process modelling, and the development of numerical simulation tools are
indispensable to optimize the mechanical strength of the weld joint. The task is more difficult in the
case of highly heterogeneous and anisotropic composite materials. Moreover, the laser transmission is
still difficult in the case of opaque or semi-transparent media such as natural fibre reinforced
thermoplastic composites. The thermal and optical properties of composites depend on the properties
and morphology of the constituents such as fibres and polymer, which can affect the transmission
spectrum in the infrared range. The absorption and refraction of laser ray propagation in the composite
materials lead to a reduction of the transmitted energy arriving at the weld interface, which directly
influences the quality of the weld and its mechanical performance.
In this dissertation, the effect of absorption and diffusion phenomena on the development of temperature
field at the weld interface is analysed numerically and experimentally. Considering the fibre orientation,
shape, length and volume fraction, numerical 3D geometries representing composite materials are
generated to simulate the propagation of laser rays with “Ray tracing” algorithm. Numerical models to
estimate the strength of weld are presented while considering the influence of welding parameters (such
as laser power, feeding speed and focus position), material properties and molecular interdiffusion at
the weld interface. The weld bonding strength is measured by mechanical tests and their results are
compared with numerical modelling results.
Keywords: Laser transmission welding, natural fibre reinforced composites, ray tracing, molecular
interdiffusion
Résume
Le soudage par transmission laser des thermoplastiques nécessite l'optimisation de l'adhérence
interfaciale au niveau du joint de soudure. A cet égard, la modélisation du procédé et le développement
d'outils de simulation numérique sont indispensables pour optimiser la résistance mécanique du joint
de soudure. La tâche est plus difficile dans le cas de matériaux composites très hétérogènes et
anisotropes. De plus, la transmission laser est encore difficile dans le cas de milieux opaques ou semi-
transparents tels que les composites thermoplastiques renforcés de fibres naturelles. Les propriétés
thermiques et optiques des composites dépendent des propriétés et de la morphologie des constituants
tels que les fibres et le polymère, qui peuvent affecter le spectre de transmission dans le domaine
infrarouge. L'absorption et la réfraction de la propagation du rayon laser dans les matériaux composites
conduisent à une réduction de l'énergie transmise arrivant à l'interface de soudure, ce qui influence
directement la qualité de la soudure et ses performances mécaniques.
Dans cette thèse, l'effet des phénomènes d'absorption et de diffusion sur le développement du champ
de température à l'interface de la soudure est analysé numériquement et expérimentalement. Compte
tenu de l'orientation, de la forme, de la longueur et de la fraction volumique des fibres, des géométries
numériques 3D représentant les matériaux composites sont générées pour simuler la propagation des
rayons laser avec l'algorithme de "Ray tracing". Des modèles numériques pour estimer la résistance de
la soudure sont présentés tout en tenant compte de l'influence des paramètres de soudage (tels que la
puissance du laser, la vitesse d'alimentation et la position du foyer), les propriétés du matériau et
l'interdiffusion moléculaire à l'interface de la soudure. La résistance de la soudure est mesurée par des
essais mécaniques et leurs résultats sont comparés aux résultats de la modélisation numérique.
Mots clés : Soudage par transmission laser, matériaux composites, ray tracing, interdiffusion
moléculaire